Here's how to support cuda11/pytorch-1.10

[ajtao]

I have an Ampere-class GPU and so i needed to get this code working with cuda-11. I think a lot of people might benefit from this, so i'll just share it here

Here are the steps:

conda create -n pedestron python=3.8 anaconda conda activate pedestron pip install mmcv==0.2.10 conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch -y

not sure these are needed ...

pip install cython pip install numpy pip install instaboostfast pip install git+https://github.com/cocodataset/panopticapi.git pip install git+https://github.com/lvis-dataset/lvis-api.git pip install albumentations>=0.3.2 --no-binary imgaug,albumentations
pip install terminaltables

modify Pedestron code to change all AT_CHECK to TORCH_CHECK

python setup.py develop

After this i was able to run demo.py successfully.

[hasanirtiza]

This is cool, could you also confirm if you can train using the same environment. I will add this to the installation file then.