DeepLabV3 is a semantic segmentation architecture that improves upon DeepLabV2 with several modifications. To handle the problem of segmenting objects at multiple scales, modules are designed which employ atrous convolution in cascade or in parallel to capture multi-scale context by adopting multiple atrous rates.
pip3 install 'scipy' 'matplotlib' 'pycocotools' 'opencv-python' 'easydict' 'tqdm'
Go to visit COCO official website, then select the COCO dataset you want to download.
Take coco2017 dataset as an example, specify /path/to/coco2017 to your COCO path in later training process, the unzipped dataset path structure sholud look like:
coco2017
├── annotations
│ ├── instances_train2017.json
│ ├── instances_val2017.json
│ └── ...
├── train2017
│ ├── 000000000009.jpg
│ ├── 000000000025.jpg
│ └── ...
├── val2017
│ ├── 000000000139.jpg
│ ├── 000000000285.jpg
│ └── ...
├── train2017.txt
├── val2017.txt
└── ...bash train_deeplabv3_r50_dist.sh --data-path /path/to/coco2017/ --dataset cocoRef: torchvision