Environment

[LIYangggggg]

When I want to run train.py in the tools directory, it prompts me that mmcv=1.0.5 is required. Does the training of this program only work in this environment?

[LIYangggggg]

What's more, is there a trained model for us to test? Thanks

[tcourat]

When I want to run train.py in the tools directory, it prompts me that mmcv=1.0.5 is required. Does the training of this program only work in this environment?

I had the same issue. I changed the line
mmcv_maximum_version = '1.0.5'
to
mmcv_maximum_version = '1.7.0' (1.7.0 is my version)
in ./mmdet/init.py

and I am now able do the training of the model with ./tools/train.py. However I do not know is this breaks something in the code, I did not tested everything.

[LIYangggggg]

You need change your environment to mmcv='1.0.5', so that you can run ./tools/train.py like me

[likaiucas]

when install this project, we should run "pip install -e ." in an mmdetection download from openmmlab or in this repo "BONAI"? COZ, my device is 3090, it only supports CUDA>=11.1, But this code is edited from mmdetection=2.3.0 which need pytorch<=1.5