bevfusion

bevfusion copied to clipboard

Thanks a lot for your work. I tried to test the pretrained model on 2 3090 gpus and encountered with some problems

1. I run torchpack dist-run -np 2 python tools/test.py configs/nuscenes/seg/fusion-bev256d2-lss.yaml pretrained/bevfusion-seg.pth --eval map and the program stuck in the last like this

2. I alse tried to visualize the result. I run torchpack dist-run -np 2 python tools/visualize.py configs/nuscenes/seg/fusion-bev256d2-lss.yaml --model pred --checkpoint pretrained/bevfusion-seg.pth --out-dir result/visualize , then the error was N > 0 assert faild. CUDA kernel launch blocks must be positive, but got N= 0, I followed other issues to checkout the version of mmcv mmcv-full are 1.4.0 and when run test.py I did not meet the same problem. Then I run the visualize code on detection model the following error appears, it seems like the authors added radar data in to the pipeline, how can I used the downloaded pretrained model?

Could you try running the evaluation on a single GPU to see if the same issue still persists? Thank you.