search

microsoft / scene_graph_benchmark

image scene graph generation benchmark
MIT License
380 stars 86 forks source link

Couldn't load custom C++ ops. #60

Closed QC-LY closed 2 years ago

QC-LY commented 2 years ago

I installed the environment as required, pytorch = = 1.7.1 torchvision = = 0.8.2 torchaudio = = 0.7.2 cudatoolkit = 10.1, but when I run test sg net.py, an error occurred: \ Runtime error occurred in Image Ids: 0,1 Couldn't load custom C++ ops. This can happen if your PyTorch and torchvision versions are incompatible, or if you had errors while compiling torchvision from source. For further information on the compatible versions, check https://github.com/pytorch/vision#installation for the compatibility matrix. Please check your PyTorch version with torch.__version__ and your torchvision version with torchvision.__version__ and verify if they are compatible, and if not please reinstall torchvision so that it matches your PyTorch install. \ Can anyone help me? \ Thanks

vxzcorporation commented 2 years ago

Hi :) Hello from Korea. I encountered exactly same problem. We are stuck. Do you have any proper solution for this problem? (I hope you closed this issue because you found the solution for this problem.) If there is a proper solution, I hope you can share :)

KatieGou commented 2 years ago

Hi I have the same problem. Could you share the solution if any?

KatieGou commented 2 years ago

Hi :) Hello from Korea. I encountered exactly same problem. We are stuck. Do you have any proper solution for this problem? (I hope you closed this issue because you found the solution for this problem.) If there is a proper solution, I hope you can share :)

Hi, I solved it by uninstalling and reinstall torchvision and torch. Hope it helps!

DavidHuji commented 2 years ago

Hi, I have the same issue, it did not help to reinstall but I guess it is because I have a few Cuda versions installed and the default one is 11.5 rather than 11. You can try to specify the version by something like "export CUDA_HOME=/pathTo/cuda-11 " and then run again setup.py .

Pter61 commented 2 years ago

I fixed this problem by uninstall the torchvision. You don't need to follow the install.md, just follow this steps: pip uninstall torchvision conda install torchvision Then you can fix this problem.