doi
stringlengths 9
10
| chunk-id
stringlengths 1
4
| chunk
stringlengths 1
1.57k
| id
stringlengths 9
10
| title
stringlengths 10
127
| summary
stringlengths 581
1.92k
| source
stringlengths 30
31
| authors
sequence | categories
sequence | comment
stringclasses 28
values | journal_ref
stringclasses 1
value | primary_category
stringclasses 6
values | published
stringlengths 8
8
| updated
stringlengths 8
8
| references
list |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2102.05095 | 71 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 72 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 73 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 74 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 75 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 76 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 77 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit> <latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 78 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 79 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 80 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 81 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 82 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 83 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 84 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 85 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit> <latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 86 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 87 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 88 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 89 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 90 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 91 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 92 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 93 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit> | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 94 | Space Attention (S) Joint Space-Time Attention (ST)Divided Space-Time Attention (T+S) <latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 95 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 96 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 97 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 98 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 99 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 100 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 101 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 102 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit>Time | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 103 | Att.Width Att. <latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 104 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 105 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 106 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 107 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 108 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 109 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 110 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 111 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit> | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 112 | MLP <latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 113 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 114 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 115 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 116 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 117 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 118 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 119 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 120 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit>Height | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 121 | Att. <latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 122 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 123 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 124 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 125 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 126 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 127 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 128 | sha1_base64="N4ztu7rA6GcWPA6nADZ8tMOcvqE=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKeyKoMegF48RzAOSJcxOZpMxszPLTK8QQv7BiwdFvPo/3vwbJ8keNLGgoajqprsrSqWw6PvfXmFtfWNzq7hd2tnd2z8oHx41rc4M4w2mpTbtiFouheINFCh5OzWcJpHkrWh0O/NbT9xYodUDjlMeJnSgRCwYRSc1uzqVme2VK37Vn4OskiAnFchR75W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 129 | 5W/un3NsoQrZJJa2wn8FMMJNSiY5NNSN7M8pWxEB7zjqKIJt+Fkfu2UnDmlT2JtXCkkc/X3xIQm1o6TyHUmFId22ZuJ/3mdDOPrcCJUmiFXbLEoziRBTWavk74wnKEcO0KZEe5WwobUUIYuoJILIVh+eZU0L6qBXw3uLyu1mzyOIpzAKZxDAFdQgzuoQwMYPMIzvMKbp70X7937WLQWvHzmGP7A+/wBz/6PRQ==</latexit> | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 130 | Axial Attention (T+W+H)Sparse Local Global Attention (L+G)MLPLocal Att.Global Att.MLPz(`) | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 131 | <latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 132 | sha1_base64="vr9M27hwkmJn9HyewcyBFvyrmCg=">AAAB+XicbVBNS8NAEN3Ur1q/oh69LBahXkoigh6LXjxWsB/QxLLZTtqlm03Y3RRqyD/x4kERr/4Tb/4bt20O2vpg4PHeDDPzgoQzpR3n2yqtrW9sbpW3Kzu7e/sH9uFRW8WppNCiMY9lNyAKOBPQ0kxz6CYSSBRw6ATj25nfmYBULBYPepqAH5GhYCGjRBupb9uZF4T4KX/Mah5wfp737apTd+bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRml | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 133 | +bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRmlENe8VIFCaFjMoSeoYJEoPxsfnmOz4wywGEsTQmN5+rviYxESk2jwHRGRI/UsjcT//N6qQ6v/YyJJNUg6GJRmHKsYzyLAQ+YBKr51BBCJTO3YjoiklBtwqqYENzll1dJ+6LuOnX3/rLauCniKKMTdIpqyEVXqIHuUBO1EEUT9Ixe0ZuVWS/Wu/WxaC1Zxcwx+gPr8wcJUZNB</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 134 | sha1_base64="vr9M27hwkmJn9HyewcyBFvyrmCg=">AAAB+XicbVBNS8NAEN3Ur1q/oh69LBahXkoigh6LXjxWsB/QxLLZTtqlm03Y3RRqyD/x4kERr/4Tb/4bt20O2vpg4PHeDDPzgoQzpR3n2yqtrW9sbpW3Kzu7e/sH9uFRW8WppNCiMY9lNyAKOBPQ0kxz6CYSSBRw6ATj25nfmYBULBYPepqAH5GhYCGjRBupb9uZF4T4KX/Mah5wfp737apTd+bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRml | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 135 | +bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRmlENe8VIFCaFjMoSeoYJEoPxsfnmOz4wywGEsTQmN5+rviYxESk2jwHRGRI/UsjcT//N6qQ6v/YyJJNUg6GJRmHKsYzyLAQ+YBKr51BBCJTO3YjoiklBtwqqYENzll1dJ+6LuOnX3/rLauCniKKMTdIpqyEVXqIHuUBO1EEUT9Ixe0ZuVWS/Wu/WxaC1Zxcwx+gPr8wcJUZNB</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 136 | sha1_base64="vr9M27hwkmJn9HyewcyBFvyrmCg=">AAAB+XicbVBNS8NAEN3Ur1q/oh69LBahXkoigh6LXjxWsB/QxLLZTtqlm03Y3RRqyD/x4kERr/4Tb/4bt20O2vpg4PHeDDPzgoQzpR3n2yqtrW9sbpW3Kzu7e/sH9uFRW8WppNCiMY9lNyAKOBPQ0kxz6CYSSBRw6ATj25nfmYBULBYPepqAH5GhYCGjRBupb9uZF4T4KX/Mah5wfp737apTd+bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRml | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 137 | +bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRmlENe8VIFCaFjMoSeoYJEoPxsfnmOz4wywGEsTQmN5+rviYxESk2jwHRGRI/UsjcT//N6qQ6v/YyJJNUg6GJRmHKsYzyLAQ+YBKr51BBCJTO3YjoiklBtwqqYENzll1dJ+6LuOnX3/rLauCniKKMTdIpqyEVXqIHuUBO1EEUT9Ixe0ZuVWS/Wu/WxaC1Zxcwx+gPr8wcJUZNB</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 138 | sha1_base64="vr9M27hwkmJn9HyewcyBFvyrmCg=">AAAB+XicbVBNS8NAEN3Ur1q/oh69LBahXkoigh6LXjxWsB/QxLLZTtqlm03Y3RRqyD/x4kERr/4Tb/4bt20O2vpg4PHeDDPzgoQzpR3n2yqtrW9sbpW3Kzu7e/sH9uFRW8WppNCiMY9lNyAKOBPQ0kxz6CYSSBRw6ATj25nfmYBULBYPepqAH5GhYCGjRBupb9uZF4T4KX/Mah5wfp737apTd+bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRml | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 139 | +bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRmlENe8VIFCaFjMoSeoYJEoPxsfnmOz4wywGEsTQmN5+rviYxESk2jwHRGRI/UsjcT//N6qQ6v/YyJJNUg6GJRmHKsYzyLAQ+YBKr51BBCJTO3YjoiklBtwqqYENzll1dJ+6LuOnX3/rLauCniKKMTdIpqyEVXqIHuUBO1EEUT9Ixe0ZuVWS/Wu/WxaC1Zxcwx+gPr8wcJUZNB</latexit>z(` 1) | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 140 | <latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 141 | sha1_base64="PqFOkq34IvzyoSmk0/rjIYa/Lb0=">AAAB+3icbVBNS8NAEN3Ur1q/Yj16WSxCPVgSEfRY9OKxgv2ANpbNdtIu3WzC7kasIX/FiwdFvPpHvPlv3LY5aOuDgcd7M8zM82POlHacb6uwsrq2vlHcLG1t7+zu2fvllooSSaFJIx7Jjk8UcCagqZnm0IklkNDn0PbH11O//QBSsUjc6UkMXkiGggWMEm2kvl1Oe36An7L7tNoDzk/dk6xvV5yaMwNeJm5OKihHo29/9QYRTUIQmnKiVNd1Yu2lR | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 142 | 5yaMwNeJm5OKihHo29/9QYRTUIQmnKiVNd1Yu2lRGpGOWSlXqIgJnRMhtA1VJAQlJfObs/wsVEGOIikKaHxTP09kZJQqUnom86Q6JFa9Kbif1430cGllzIRJxoEnS8KEo51hKdB4AGTQDWfGEKoZOZWTEdEEqpNXCUTgrv48jJpndVcp+benlfqV3kcRXSIjlAVuegC1dENaqAmougRPaNX9GZl1ov1bn3MWwtWPnOA/sD6/AHtlpOz</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 143 | sha1_base64="PqFOkq34IvzyoSmk0/rjIYa/Lb0=">AAAB+3icbVBNS8NAEN3Ur1q/Yj16WSxCPVgSEfRY9OKxgv2ANpbNdtIu3WzC7kasIX/FiwdFvPpHvPlv3LY5aOuDgcd7M8zM82POlHacb6uwsrq2vlHcLG1t7+zu2fvllooSSaFJIx7Jjk8UcCagqZnm0IklkNDn0PbH11O//QBSsUjc6UkMXkiGggWMEm2kvl1Oe36An7L7tNoDzk/dk6xvV5yaMwNeJm5OKihHo29/9QYRTUIQmnKiVNd1Yu2lR | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 144 | 5yaMwNeJm5OKihHo29/9QYRTUIQmnKiVNd1Yu2lRGpGOWSlXqIgJnRMhtA1VJAQlJfObs/wsVEGOIikKaHxTP09kZJQqUnom86Q6JFa9Kbif1430cGllzIRJxoEnS8KEo51hKdB4AGTQDWfGEKoZOZWTEdEEqpNXCUTgrv48jJpndVcp+benlfqV3kcRXSIjlAVuegC1dENaqAmougRPaNX9GZl1ov1bn3MWwtWPnOA/sD6/AHtlpOz</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 145 | sha1_base64="PqFOkq34IvzyoSmk0/rjIYa/Lb0=">AAAB+3icbVBNS8NAEN3Ur1q/Yj16WSxCPVgSEfRY9OKxgv2ANpbNdtIu3WzC7kasIX/FiwdFvPpHvPlv3LY5aOuDgcd7M8zM82POlHacb6uwsrq2vlHcLG1t7+zu2fvllooSSaFJIx7Jjk8UcCagqZnm0IklkNDn0PbH11O//QBSsUjc6UkMXkiGggWMEm2kvl1Oe36An7L7tNoDzk/dk6xvV5yaMwNeJm5OKihHo29/9QYRTUIQmnKiVNd1Yu2lR | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 146 | 5yaMwNeJm5OKihHo29/9QYRTUIQmnKiVNd1Yu2lRGpGOWSlXqIgJnRMhtA1VJAQlJfObs/wsVEGOIikKaHxTP09kZJQqUnom86Q6JFa9Kbif1430cGllzIRJxoEnS8KEo51hKdB4AGTQDWfGEKoZOZWTEdEEqpNXCUTgrv48jJpndVcp+benlfqV3kcRXSIjlAVuegC1dENaqAmougRPaNX9GZl1ov1bn3MWwtWPnOA/sD6/AHtlpOz</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 147 | sha1_base64="PqFOkq34IvzyoSmk0/rjIYa/Lb0=">AAAB+3icbVBNS8NAEN3Ur1q/Yj16WSxCPVgSEfRY9OKxgv2ANpbNdtIu3WzC7kasIX/FiwdFvPpHvPlv3LY5aOuDgcd7M8zM82POlHacb6uwsrq2vlHcLG1t7+zu2fvllooSSaFJIx7Jjk8UcCagqZnm0IklkNDn0PbH11O//QBSsUjc6UkMXkiGggWMEm2kvl1Oe36An7L7tNoDzk/dk6xvV5yaMwNeJm5OKihHo29/9QYRTUIQmnKiVNd1Yu2lR | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 148 | 5yaMwNeJm5OKihHo29/9QYRTUIQmnKiVNd1Yu2lRGpGOWSlXqIgJnRMhtA1VJAQlJfObs/wsVEGOIikKaHxTP09kZJQqUnom86Q6JFa9Kbif1430cGllzIRJxoEnS8KEo51hKdB4AGTQDWfGEKoZOZWTEdEEqpNXCUTgrv48jJpndVcp+benlfqV3kcRXSIjlAVuegC1dENaqAmougRPaNX9GZl1ov1bn3MWwtWPnOA/sD6/AHtlpOz</latexit> | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 149 | z(`) | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 150 | <latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 151 | sha1_base64="vr9M27hwkmJn9HyewcyBFvyrmCg=">AAAB+XicbVBNS8NAEN3Ur1q/oh69LBahXkoigh6LXjxWsB/QxLLZTtqlm03Y3RRqyD/x4kERr/4Tb/4bt20O2vpg4PHeDDPzgoQzpR3n2yqtrW9sbpW3Kzu7e/sH9uFRW8WppNCiMY9lNyAKOBPQ0kxz6CYSSBRw6ATj25nfmYBULBYPepqAH5GhYCGjRBupb9uZF4T4KX/Mah5wfp737apTd+bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRml | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 152 | +bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRmlENe8VIFCaFjMoSeoYJEoPxsfnmOz4wywGEsTQmN5+rviYxESk2jwHRGRI/UsjcT//N6qQ6v/YyJJNUg6GJRmHKsYzyLAQ+YBKr51BBCJTO3YjoiklBtwqqYENzll1dJ+6LuOnX3/rLauCniKKMTdIpqyEVXqIHuUBO1EEUT9Ixe0ZuVWS/Wu/WxaC1Zxcwx+gPr8wcJUZNB</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 153 | sha1_base64="vr9M27hwkmJn9HyewcyBFvyrmCg=">AAAB+XicbVBNS8NAEN3Ur1q/oh69LBahXkoigh6LXjxWsB/QxLLZTtqlm03Y3RRqyD/x4kERr/4Tb/4bt20O2vpg4PHeDDPzgoQzpR3n2yqtrW9sbpW3Kzu7e/sH9uFRW8WppNCiMY9lNyAKOBPQ0kxz6CYSSBRw6ATj25nfmYBULBYPepqAH5GhYCGjRBupb9uZF4T4KX/Mah5wfp737apTd+bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRml | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 154 | +bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRmlENe8VIFCaFjMoSeoYJEoPxsfnmOz4wywGEsTQmN5+rviYxESk2jwHRGRI/UsjcT//N6qQ6v/YyJJNUg6GJRmHKsYzyLAQ+YBKr51BBCJTO3YjoiklBtwqqYENzll1dJ+6LuOnX3/rLauCniKKMTdIpqyEVXqIHuUBO1EEUT9Ixe0ZuVWS/Wu/WxaC1Zxcwx+gPr8wcJUZNB</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 155 | sha1_base64="vr9M27hwkmJn9HyewcyBFvyrmCg=">AAAB+XicbVBNS8NAEN3Ur1q/oh69LBahXkoigh6LXjxWsB/QxLLZTtqlm03Y3RRqyD/x4kERr/4Tb/4bt20O2vpg4PHeDDPzgoQzpR3n2yqtrW9sbpW3Kzu7e/sH9uFRW8WppNCiMY9lNyAKOBPQ0kxz6CYSSBRw6ATj25nfmYBULBYPepqAH5GhYCGjRBupb9uZF4T4KX/Mah5wfp737apTd+bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRml | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 156 | +bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRmlENe8VIFCaFjMoSeoYJEoPxsfnmOz4wywGEsTQmN5+rviYxESk2jwHRGRI/UsjcT//N6qQ6v/YyJJNUg6GJRmHKsYzyLAQ+YBKr51BBCJTO3YjoiklBtwqqYENzll1dJ+6LuOnX3/rLauCniKKMTdIpqyEVXqIHuUBO1EEUT9Ixe0ZuVWS/Wu/WxaC1Zxcwx+gPr8wcJUZNB</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 157 | sha1_base64="vr9M27hwkmJn9HyewcyBFvyrmCg=">AAAB+XicbVBNS8NAEN3Ur1q/oh69LBahXkoigh6LXjxWsB/QxLLZTtqlm03Y3RRqyD/x4kERr/4Tb/4bt20O2vpg4PHeDDPzgoQzpR3n2yqtrW9sbpW3Kzu7e/sH9uFRW8WppNCiMY9lNyAKOBPQ0kxz6CYSSBRw6ATj25nfmYBULBYPepqAH5GhYCGjRBupb9uZF4T4KX/Mah5wfp737apTd+bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRml | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 158 | +bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRmlENe8VIFCaFjMoSeoYJEoPxsfnmOz4wywGEsTQmN5+rviYxESk2jwHRGRI/UsjcT//N6qQ6v/YyJJNUg6GJRmHKsYzyLAQ+YBKr51BBCJTO3YjoiklBtwqqYENzll1dJ+6LuOnX3/rLauCniKKMTdIpqyEVXqIHuUBO1EEUT9Ixe0ZuVWS/Wu/WxaC1Zxcwx+gPr8wcJUZNB</latexit>z(`) | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 159 | <latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 160 | sha1_base64="vr9M27hwkmJn9HyewcyBFvyrmCg=">AAAB+XicbVBNS8NAEN3Ur1q/oh69LBahXkoigh6LXjxWsB/QxLLZTtqlm03Y3RRqyD/x4kERr/4Tb/4bt20O2vpg4PHeDDPzgoQzpR3n2yqtrW9sbpW3Kzu7e/sH9uFRW8WppNCiMY9lNyAKOBPQ0kxz6CYSSBRw6ATj25nfmYBULBYPepqAH5GhYCGjRBupb9uZF4T4KX/Mah5wfp737apTd+bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRml | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 161 | +bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRmlENe8VIFCaFjMoSeoYJEoPxsfnmOz4wywGEsTQmN5+rviYxESk2jwHRGRI/UsjcT//N6qQ6v/YyJJNUg6GJRmHKsYzyLAQ+YBKr51BBCJTO3YjoiklBtwqqYENzll1dJ+6LuOnX3/rLauCniKKMTdIpqyEVXqIHuUBO1EEUT9Ixe0ZuVWS/Wu/WxaC1Zxcwx+gPr8wcJUZNB</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 162 | sha1_base64="vr9M27hwkmJn9HyewcyBFvyrmCg=">AAAB+XicbVBNS8NAEN3Ur1q/oh69LBahXkoigh6LXjxWsB/QxLLZTtqlm03Y3RRqyD/x4kERr/4Tb/4bt20O2vpg4PHeDDPzgoQzpR3n2yqtrW9sbpW3Kzu7e/sH9uFRW8WppNCiMY9lNyAKOBPQ0kxz6CYSSBRw6ATj25nfmYBULBYPepqAH5GhYCGjRBupb9uZF4T4KX/Mah5wfp737apTd+bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRml | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 163 | +bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRmlENe8VIFCaFjMoSeoYJEoPxsfnmOz4wywGEsTQmN5+rviYxESk2jwHRGRI/UsjcT//N6qQ6v/YyJJNUg6GJRmHKsYzyLAQ+YBKr51BBCJTO3YjoiklBtwqqYENzll1dJ+6LuOnX3/rLauCniKKMTdIpqyEVXqIHuUBO1EEUT9Ixe0ZuVWS/Wu/WxaC1Zxcwx+gPr8wcJUZNB</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 164 | sha1_base64="vr9M27hwkmJn9HyewcyBFvyrmCg=">AAAB+XicbVBNS8NAEN3Ur1q/oh69LBahXkoigh6LXjxWsB/QxLLZTtqlm03Y3RRqyD/x4kERr/4Tb/4bt20O2vpg4PHeDDPzgoQzpR3n2yqtrW9sbpW3Kzu7e/sH9uFRW8WppNCiMY9lNyAKOBPQ0kxz6CYSSBRw6ATj25nfmYBULBYPepqAH5GhYCGjRBupb9uZF4T4KX/Mah5wfp737apTd+bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRml | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 165 | +bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRmlENe8VIFCaFjMoSeoYJEoPxsfnmOz4wywGEsTQmN5+rviYxESk2jwHRGRI/UsjcT//N6qQ6v/YyJJNUg6GJRmHKsYzyLAQ+YBKr51BBCJTO3YjoiklBtwqqYENzll1dJ+6LuOnX3/rLauCniKKMTdIpqyEVXqIHuUBO1EEUT9Ixe0ZuVWS/Wu/WxaC1Zxcwx+gPr8wcJUZNB</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 166 | sha1_base64="vr9M27hwkmJn9HyewcyBFvyrmCg=">AAAB+XicbVBNS8NAEN3Ur1q/oh69LBahXkoigh6LXjxWsB/QxLLZTtqlm03Y3RRqyD/x4kERr/4Tb/4bt20O2vpg4PHeDDPzgoQzpR3n2yqtrW9sbpW3Kzu7e/sH9uFRW8WppNCiMY9lNyAKOBPQ0kxz6CYSSBRw6ATj25nfmYBULBYPepqAH5GhYCGjRBupb9uZF4T4KX/Mah5wfp737apTd+bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRml | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 167 | +bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRmlENe8VIFCaFjMoSeoYJEoPxsfnmOz4wywGEsTQmN5+rviYxESk2jwHRGRI/UsjcT//N6qQ6v/YyJJNUg6GJRmHKsYzyLAQ+YBKr51BBCJTO3YjoiklBtwqqYENzll1dJ+6LuOnX3/rLauCniKKMTdIpqyEVXqIHuUBO1EEUT9Ixe0ZuVWS/Wu/WxaC1Zxcwx+gPr8wcJUZNB</latexit>z(`) | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 168 | <latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 169 | sha1_base64="vr9M27hwkmJn9HyewcyBFvyrmCg=">AAAB+XicbVBNS8NAEN3Ur1q/oh69LBahXkoigh6LXjxWsB/QxLLZTtqlm03Y3RRqyD/x4kERr/4Tb/4bt20O2vpg4PHeDDPzgoQzpR3n2yqtrW9sbpW3Kzu7e/sH9uFRW8WppNCiMY9lNyAKOBPQ0kxz6CYSSBRw6ATj25nfmYBULBYPepqAH5GhYCGjRBupb9uZF4T4KX/Mah5wfp737apTd+bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRml | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
2102.05095 | 170 | +bAq8QtSBUVaPbtL28Q0zQCoSknSvVcJ9F+RqRmlENe8VIFCaFjMoSeoYJEoPxsfnmOz4wywGEsTQmN5+rviYxESk2jwHRGRI/UsjcT//N6qQ6v/YyJJNUg6GJRmHKsYzyLAQ+YBKr51BBCJTO3YjoiklBtwqqYENzll1dJ+6LuOnX3/rLauCniKKMTdIpqyEVXqIHuUBO1EEUT9Ixe0ZuVWS/Wu/WxaC1Zxcwx+gPr8wcJUZNB</latexit><latexit | 2102.05095 | Is Space-Time Attention All You Need for Video Understanding? | We present a convolution-free approach to video classification built
exclusively on self-attention over space and time. Our method, named
"TimeSformer," adapts the standard Transformer architecture to video by
enabling spatiotemporal feature learning directly from a sequence of
frame-level patches. Our experimental study compares different self-attention
schemes and suggests that "divided attention," where temporal attention and
spatial attention are separately applied within each block, leads to the best
video classification accuracy among the design choices considered. Despite the
radically new design, TimeSformer achieves state-of-the-art results on several
action recognition benchmarks, including the best reported accuracy on
Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks,
our model is faster to train, it can achieve dramatically higher test
efficiency (at a small drop in accuracy), and it can also be applied to much
longer video clips (over one minute long). Code and models are available at:
https://github.com/facebookresearch/TimeSformer. | http://arxiv.org/pdf/2102.05095 | [
"Gedas Bertasius",
"Heng Wang",
"Lorenzo Torresani"
] | [
"cs.CV"
] | Accepted to ICML 2021 | null | cs.CV | 20210209 | 20210609 | [
{
"id": "2005.00928",
"title": "Quantifying Attention Flow in Transformers",
"authors": "Abnar, S. and Zuidema, W.",
"year": "2020"
},
{
"id": "1607.06450",
"title": "Layer Normalization",
"authors": "Lei Jimmy Ba, Jamie Ryan Kiros, Geoffrey E. Hinton",
"year": "2016"
},
{
"id": "1904.09925",
"title": "Attention Augmented Convolutional Networks",
"authors": "Bello, I., Zoph, B., Le, Q., Vaswani, A., Shlens, J.",
"year": "2019"
},
{
"id": "1912.04573",
"title": "Classifying, Segmenting, and Tracking Object Instances in Video with Mask Propagation",
"authors": "Bertasius, G. and Torresani, L.",
"year": "2020"
},
{
"id": "2005.14165",
"title": "Language Models are Few-Shot Learners",
"authors": "Tom B. Brown, et al.",
"year": "2020"
}
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.