please refer to the requirements.txt file.
recommand to use conda virtual env.
note that some requirements are not so strict.
pip install -r requirements.txt
pytorch 1.7.0
mmcv-full 1.3.16
cuda 11.4
apex(recommand)
cv2
pip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/cu110/torch1.7.0/index.html
See here for different versions of MMCV compatible to different PyTorch and CUDA versions.
We apply the backbone of Video-Swin Transformer as the feature extractor. The repo mmaction has provied flexible api to use it. Please refer to it.
git clone https://github.com/open-mmlab/mmaction2.git
cd mmaction2
Install build requirements and then install MMAction2.
pip install -r requirements/build.txt
pip install -v -e . # or "python setup.py develop"
pip install tqdm tensorboardX timm einops