--- library_name: pytorch license: other tags: - android pipeline_tag: image-segmentation --- ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/deeplabv3_plus_mobilenet/web-assets/model_demo.png) # DeepLabV3-Plus-MobileNet: Optimized for Mobile Deployment ## Deep Convolutional Neural Network model for semantic segmentation DeepLabV3 is designed for semantic segmentation at multiple scales, trained on the various datasets. It uses MobileNet as a backbone. This model is an implementation of DeepLabV3-Plus-MobileNet found [here](https://github.com/jfzhang95/pytorch-deeplab-xception). This repository provides scripts to run DeepLabV3-Plus-MobileNet on Qualcomm® devices. More details on model performance across various devices, can be found [here](https://aihub.qualcomm.com/models/deeplabv3_plus_mobilenet). ### Model Details - **Model Type:** Model_use_case.semantic_segmentation - **Model Stats:** - Model checkpoint: VOC2012 - Input resolution: 513x513 - Number of output classes: 21 - Number of parameters: 5.80M - Model size (float): 22.2 MB - Model size (w8a16): 6.67 MB | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model |---|---|---|---|---|---|---|---|---| | DeepLabV3-Plus-MobileNet | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 62.603 ms | 0 - 28 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) | | DeepLabV3-Plus-MobileNet | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 58.62 ms | 2 - 32 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) | | DeepLabV3-Plus-MobileNet | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 19.174 ms | 0 - 45 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) | | DeepLabV3-Plus-MobileNet | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 21.272 ms | 3 - 49 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) | | DeepLabV3-Plus-MobileNet | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 13.008 ms | 0 - 12 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) | | DeepLabV3-Plus-MobileNet | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 11.24 ms | 3 - 12 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) | | DeepLabV3-Plus-MobileNet | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 19.529 ms | 0 - 29 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) | | DeepLabV3-Plus-MobileNet | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 17.545 ms | 2 - 33 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) | | DeepLabV3-Plus-MobileNet | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 62.603 ms | 0 - 28 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) | | DeepLabV3-Plus-MobileNet | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 58.62 ms | 2 - 32 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) | | DeepLabV3-Plus-MobileNet | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 12.999 ms | 0 - 13 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) | | DeepLabV3-Plus-MobileNet | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 11.222 ms | 3 - 15 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) | | DeepLabV3-Plus-MobileNet | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 22.216 ms | 0 - 30 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) | | DeepLabV3-Plus-MobileNet | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 19.839 ms | 3 - 45 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) | | DeepLabV3-Plus-MobileNet | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 13.023 ms | 0 - 11 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) | | DeepLabV3-Plus-MobileNet | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 11.216 ms | 3 - 14 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) | | DeepLabV3-Plus-MobileNet | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 19.529 ms | 0 - 29 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) | | DeepLabV3-Plus-MobileNet | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 17.545 ms | 2 - 33 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) | | DeepLabV3-Plus-MobileNet | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 13.018 ms | 0 - 12 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) | | DeepLabV3-Plus-MobileNet | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 11.296 ms | 3 - 15 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) | | DeepLabV3-Plus-MobileNet | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 10.835 ms | 4 - 14 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.onnx) | | DeepLabV3-Plus-MobileNet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 9.086 ms | 0 - 46 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) | | DeepLabV3-Plus-MobileNet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 7.871 ms | 3 - 44 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) | | DeepLabV3-Plus-MobileNet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 7.623 ms | 2 - 43 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.onnx) | | DeepLabV3-Plus-MobileNet | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 7.193 ms | 0 - 32 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) | | DeepLabV3-Plus-MobileNet | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 7.663 ms | 3 - 40 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) | | DeepLabV3-Plus-MobileNet | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 6.245 ms | 4 - 42 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.onnx) | | DeepLabV3-Plus-MobileNet | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 13.085 ms | 23 - 23 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) | | DeepLabV3-Plus-MobileNet | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 11.965 ms | 10 - 10 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.onnx) | | DeepLabV3-Plus-MobileNet | w8a16 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 21.453 ms | 2 - 40 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) | | DeepLabV3-Plus-MobileNet | w8a16 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 11.63 ms | 2 - 53 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) | | DeepLabV3-Plus-MobileNet | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 8.315 ms | 0 - 16 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) | | DeepLabV3-Plus-MobileNet | w8a16 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 9.071 ms | 2 - 40 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) | | DeepLabV3-Plus-MobileNet | w8a16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 44.053 ms | 2 - 91 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) | | DeepLabV3-Plus-MobileNet | w8a16 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 21.453 ms | 2 - 40 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) | | DeepLabV3-Plus-MobileNet | w8a16 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 8.366 ms | 2 - 16 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) | | DeepLabV3-Plus-MobileNet | w8a16 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 12.266 ms | 2 - 43 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) | | DeepLabV3-Plus-MobileNet | w8a16 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 8.337 ms | 2 - 15 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) | | DeepLabV3-Plus-MobileNet | w8a16 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 9.071 ms | 2 - 40 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) | | DeepLabV3-Plus-MobileNet | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 8.367 ms | 2 - 15 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) | | DeepLabV3-Plus-MobileNet | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 77.94 ms | 95 - 122 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.onnx) | | DeepLabV3-Plus-MobileNet | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 6.05 ms | 2 - 58 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) | | DeepLabV3-Plus-MobileNet | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 60.88 ms | 99 - 273 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.onnx) | | DeepLabV3-Plus-MobileNet | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 4.16 ms | 2 - 45 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) | | DeepLabV3-Plus-MobileNet | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 57.479 ms | 101 - 275 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.onnx) | | DeepLabV3-Plus-MobileNet | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 10.072 ms | 0 - 0 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) | | DeepLabV3-Plus-MobileNet | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 75.822 ms | 133 - 133 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.onnx) | | DeepLabV3-Plus-MobileNet | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 11.564 ms | 0 - 31 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) | | DeepLabV3-Plus-MobileNet | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 10.592 ms | 1 - 36 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) | | DeepLabV3-Plus-MobileNet | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 5.061 ms | 0 - 46 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) | | DeepLabV3-Plus-MobileNet | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 5.907 ms | 1 - 50 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) | | DeepLabV3-Plus-MobileNet | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 4.506 ms | 0 - 15 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) | | DeepLabV3-Plus-MobileNet | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 3.914 ms | 1 - 15 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) | | DeepLabV3-Plus-MobileNet | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 5.079 ms | 0 - 33 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) | | DeepLabV3-Plus-MobileNet | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 4.437 ms | 1 - 37 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) | | DeepLabV3-Plus-MobileNet | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | TFLITE | 16.175 ms | 0 - 42 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) | | DeepLabV3-Plus-MobileNet | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 17.824 ms | 0 - 42 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) | | DeepLabV3-Plus-MobileNet | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | TFLITE | 167.336 ms | 3 - 6 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) | | DeepLabV3-Plus-MobileNet | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 11.564 ms | 0 - 31 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) | | DeepLabV3-Plus-MobileNet | w8a8 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 10.592 ms | 1 - 36 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) | | DeepLabV3-Plus-MobileNet | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 4.503 ms | 0 - 18 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) | | DeepLabV3-Plus-MobileNet | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 3.951 ms | 1 - 15 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) | | DeepLabV3-Plus-MobileNet | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 6.861 ms | 0 - 35 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) | | DeepLabV3-Plus-MobileNet | w8a8 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 6.49 ms | 1 - 40 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) | | DeepLabV3-Plus-MobileNet | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 4.499 ms | 0 - 14 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) | | DeepLabV3-Plus-MobileNet | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 3.952 ms | 2 - 16 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) | | DeepLabV3-Plus-MobileNet | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 5.079 ms | 0 - 33 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) | | DeepLabV3-Plus-MobileNet | w8a8 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 4.437 ms | 1 - 37 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) | | DeepLabV3-Plus-MobileNet | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 4.501 ms | 0 - 16 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) | | DeepLabV3-Plus-MobileNet | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 3.923 ms | 1 - 14 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) | | DeepLabV3-Plus-MobileNet | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 72.192 ms | 68 - 143 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.onnx) | | DeepLabV3-Plus-MobileNet | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 3.244 ms | 0 - 46 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) | | DeepLabV3-Plus-MobileNet | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 2.741 ms | 1 - 51 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) | | DeepLabV3-Plus-MobileNet | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 57.992 ms | 100 - 276 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.onnx) | | DeepLabV3-Plus-MobileNet | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 2.967 ms | 0 - 37 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) | | DeepLabV3-Plus-MobileNet | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 2.458 ms | 1 - 43 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) | | DeepLabV3-Plus-MobileNet | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 57.1 ms | 101 - 276 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.onnx) | | DeepLabV3-Plus-MobileNet | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 5.114 ms | 8 - 8 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) | | DeepLabV3-Plus-MobileNet | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 67.772 ms | 133 - 133 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.onnx) | ## Installation Install the package via pip: ```bash pip install qai-hub-models ``` ## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`. With this API token, you can configure your client to run models on the cloud hosted devices. ```bash qai-hub configure --api_token API_TOKEN ``` Navigate to [docs](https://app.aihub.qualcomm.com/docs/) for more information. ## Demo off target The package contains a simple end-to-end demo that downloads pre-trained weights and runs this model on a sample input. ```bash python -m qai_hub_models.models.deeplabv3_plus_mobilenet.demo ``` The above demo runs a reference implementation of pre-processing, model inference, and post processing. **NOTE**: If you want running in a Jupyter Notebook or Google Colab like environment, please add the following to your cell (instead of the above). ``` %run -m qai_hub_models.models.deeplabv3_plus_mobilenet.demo ``` ### Run model on a cloud-hosted device In addition to the demo, you can also run the model on a cloud-hosted Qualcomm® device. This script does the following: * Performance check on-device on a cloud-hosted device * Downloads compiled assets that can be deployed on-device for Android. * Accuracy check between PyTorch and on-device outputs. ```bash python -m qai_hub_models.models.deeplabv3_plus_mobilenet.export ``` ``` Profiling Results ------------------------------------------------------------ DeepLabV3-Plus-MobileNet Device : cs_8275 (ANDROID 14) Runtime : TFLITE Estimated inference time (ms) : 62.6 Estimated peak memory usage (MB): [0, 28] Total # Ops : 101 Compute Unit(s) : npu (101 ops) gpu (0 ops) cpu (0 ops) ``` ## How does this work? This [export script](https://aihub.qualcomm.com/models/deeplabv3_plus_mobilenet/qai_hub_models/models/DeepLabV3-Plus-MobileNet/export.py) leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model on-device. Lets go through each step below in detail: Step 1: **Compile model for on-device deployment** To compile a PyTorch model for on-device deployment, we first trace the model in memory using the `jit.trace` and then call the `submit_compile_job` API. ```python import torch import qai_hub as hub from qai_hub_models.models.deeplabv3_plus_mobilenet import Model # Load the model torch_model = Model.from_pretrained() # Device device = hub.Device("Samsung Galaxy S24") # Trace model input_shape = torch_model.get_input_spec() sample_inputs = torch_model.sample_inputs() pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()]) # Compile model on a specific device compile_job = hub.submit_compile_job( model=pt_model, device=device, input_specs=torch_model.get_input_spec(), ) # Get target model to run on-device target_model = compile_job.get_target_model() ``` Step 2: **Performance profiling on cloud-hosted device** After compiling models from step 1. Models can be profiled model on-device using the `target_model`. Note that this scripts runs the model on a device automatically provisioned in the cloud. Once the job is submitted, you can navigate to a provided job URL to view a variety of on-device performance metrics. ```python profile_job = hub.submit_profile_job( model=target_model, device=device, ) ``` Step 3: **Verify on-device accuracy** To verify the accuracy of the model on-device, you can run on-device inference on sample input data on the same cloud hosted device. ```python input_data = torch_model.sample_inputs() inference_job = hub.submit_inference_job( model=target_model, device=device, inputs=input_data, ) on_device_output = inference_job.download_output_data() ``` With the output of the model, you can compute like PSNR, relative errors or spot check the output with expected output. **Note**: This on-device profiling and inference requires access to Qualcomm® AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup). ## Run demo on a cloud-hosted device You can also run the demo on-device. ```bash python -m qai_hub_models.models.deeplabv3_plus_mobilenet.demo --eval-mode on-device ``` **NOTE**: If you want running in a Jupyter Notebook or Google Colab like environment, please add the following to your cell (instead of the above). ``` %run -m qai_hub_models.models.deeplabv3_plus_mobilenet.demo -- --eval-mode on-device ``` ## Deploying compiled model to Android The models can be deployed using multiple runtimes: - TensorFlow Lite (`.tflite` export): [This tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a guide to deploy the .tflite model in an Android application. - QNN (`.so` export ): This [sample app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html) provides instructions on how to use the `.so` shared library in an Android application. ## View on Qualcomm® AI Hub Get more details on DeepLabV3-Plus-MobileNet's performance across various devices [here](https://aihub.qualcomm.com/models/deeplabv3_plus_mobilenet). Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/) ## License * The license for the original implementation of DeepLabV3-Plus-MobileNet can be found [here](https://github.com/jfzhang95/pytorch-deeplab-xception/blob/master/LICENSE). * The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf) ## References * [Rethinking Atrous Convolution for Semantic Image Segmentation](https://arxiv.org/abs/1706.05587) * [Source Model Implementation](https://github.com/jfzhang95/pytorch-deeplab-xception) ## Community * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI. * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).