Jae-Won Chung
New leaderboard prototype
b10121d
|
raw
history blame
940 Bytes

ShareGPT4Video dataset

For the image-to-video task, we sample 100 video-caption pairs from the ShareGPT4Video datset to feed to the diffusion model to generate videos.

Filtering the dataset

Download the dataset with captions and video paths.

wget https://huggingface.co/datasets/ShareGPT4Video/ShareGPT4Video/resolve/main/sharegpt4video_40k.jsonl

Sample video-caption pairs. You can adjust the NUM_SAMPLES variable in the script to change the size of the generated dataset. By default, 100 pairs will be sampled and saved as sharegpt4video_100.json.

python sample.py

Download and unzip the chunk of videos.

wget https://huggingface.co/datasets/ShareGPT4Video/ShareGPT4Video/resolve/main/zip_folder/panda/panda_videos_1.zip
unzip panda_videos_1.zip -d panda

Extract the first frame of the video and save under first_frame/.

pip install opencv-python
python extract_first_frame.py