search

evalcrafter / EvalCrafter

[CVPR 2024] EvalCrafter: Benchmarking and Evaluating Large Video Generation Models
http://evalcrafter.github.io
116 stars 7 forks source link

Warping Error Compuation issue #15

Closed CIntellifusion closed 1 week ago

CIntellifusion commented 2 weeks ago

Hi, thanks for your excellent work.

I notice you have comment this line : https://github.com/evalcrafter/EvalCrafter/blob/6dd627a8ac48ab1d4081ffc1abaca6704ea59c43/metrics/RAFT/optical_flow_scores.py#L28, which is import warp_utils.

But the warp_utils.detect_occlusion is used in this file. When I uncomment this line, it tells Fail to import Resample2D_cuda

I was wondering is it necessary to use warp_utils? or there is a work around.

Yaofang-Liu commented 1 week ago

Yes, warp_utils is needed. To use that, you should compile Resample2D_cuda beforehand. I've updated the README.md file to include:

cd ./metrics/RAFT/networks/resample2d_package python setup.py install --user Note that if you encounter this problem RuntimeError: GET was unable to find an engine to execute this computation when running the evaluation code, you may update Cuda to version Cuda 11.7 (Compatibility issue).

Very sorry for the confusion caused!

CIntellifusion commented 1 week ago

Thanks for your reply. But I found there is still a problem.

] /usr/include/c++/11/bits/std_function.h:435:145: error: parameter packs not expanded with ‘...’: 435 | function(_Functor&& __f) | ^ /usr/include/c++/11/bits/std_function.h:435:145: note: ‘_ArgTypes’ /usr/include/c++/11/bits/std_function.h:530:146: error: parameter packs not expanded with ‘...’: 530 | operator=(_Functor&& __f) | ^ /usr/include/c++/11/bits/std_function.h:530:146: note: ‘_ArgTypes’ ninja: build stopped: subcommand failed.

Do you have any other solutions or workarounds? I was wondering there might be a standard torch implementation or even naive cpu version. Thanks for your help.

Yaofang-Liu commented 1 week ago

Hi, I am not sure what problem you got, the installation for resample2d_cuda could be troublesome. You may ref to https://github.com/NVIDIA/flownet2-pytorch for installation, which also includes may solutions in its issues page.

CIntellifusion commented 1 week ago

Thanks for your reply. But I have solved this issue by specify cuda version to 11.7. export PATH=/usr/local/cuda-11.7/bin:$PATH export LD_LIBRARY_PATH=/usr/local/cuda-11.7/lib64:$LD_LIBRARY_PATH