Closed jiku100 closed 3 years ago
Could you send me more information like the env settings? It will help me investigate this issue.
Could you send me more information like the env settings? It will help me investigate this issue.
I used your docker image, So it will be the same as you.
OS: Ubuntu 18.04 CUDA: 10.0 Graphic card: gtx 1060 python: 3.6.9 torch: 1.3.0+cu100 torchvision: 0.4.1+cu100 tensorboard: 2.5.0 cython: 0.29.23 skimage: 0.17.2 cv2: 4.5.2 fvcore: 0.1.5
In my searching, Maybe fvcore could be the cause of the problem. https://github.com/facebookresearch/detectron2/issues/2652
Yes, the problem is caused by fvcore. You can try old version like 0.1.3.
Yes, the problem is caused by fvcore. You can try old version like 0.1.3.
I can solve that problem by using fvcore version 0.1.2. Thanks!! but, I get another errors. please help me..
Check the pretrained model?
Check the pretrained model?
Yes, I dowload from your Google Drive and put the model in projects/LISA/output_light/
I cannot reproduce this error in my docker. Could you recheck the pth file or download it again?
I cannot reproduce this error in my docker. Could you recheck the pth file or download it again?
Without any other changes, can I just put last_checkpoint.pth and model_final.pth in output_light?
Yes actually.
Yes actually.
Oh, After I rebuilded the setup.py files, then it works!! Thanks a lot!!
I use your docker image by nvidia-docker and follow your README to run demo.py but i got "cannot import name '_get_trace_graph' error" in running "python demo.py --input ./demo/web-shadow0573.jpg --output ./ --config ./config/LISA_101_FPN_3x_demo.yaml"
How can i solve this?