search

radekd91 / emoca

Official repository accompanying a CVPR 2022 paper EMOCA: Emotion Driven Monocular Face Capture And Animation. EMOCA takes a single image of a face as input and produces a 3D reconstruction. EMOCA sets the new standard on reconstructing highly emotional images in-the-wild
https://emoca.is.tue.mpg.de/
Other
668 stars 88 forks source link

The video file '/assets/data/EMOCA_test_example_data/videos/82-25-854x480_affwild2.mp4' is corrupted. Skipping it. #75

Open swerizwan opened 5 months ago

swerizwan commented 5 months ago

I don't know why I am getting that error, The path is correct and video is also fine.

The video file '/assets/data/EMOCA_test_example_data/videos/82-25-854x480_affwild2.mp4' is corrupted. Skipping it.

Processing dataset 0%| | 0/1 [00:00<?, ?it/s]/assets/data/EMOCA_test_example_data/videos/82-25-854x480_affwild2.mp4 The video file '/assets/data/EMOCA_test_example_data/videos/82-25-854x480_affwild2.mp4' is corrupted. Skipping it. 100%|███████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00, 55.22it/s] Traceback (most recent call last): File "gdl_apps/EMOCA/demos/test_emoca_on_video.py", line 172, in main() File "gdl_apps/EMOCA/demos/test_emoca_on_video.py", line 167, in main reconstruct_video(args) File "gdl_apps/EMOCA/demos/test_emoca_on_video.py", line 62, in reconstruct_video dm.prepare_data() File "/home/abbas/anaconda3/envs/emoo36/lib/python3.8/site-packages/pytorch_lightning/core/datamodule.py", line 428, in wrapped_fn fn(*args, **kwargs) File "/home/abbas/emoca/gdl/datasets/FaceVideoDataModule.py", line 3192, in prepare_data self._gather_data(exist_ok=True) File "/home/abbas/emoca/gdl/datasets/FaceVideoDataModule.py", line 3214, in _gather_data self._gather_video_metadata() File "/home/abbas/emoca/gdl/datasets/FaceVideoDataModule.py", line 1570, in _gather_video_metadata del self.annotation_list[vi] IndexError: list assignment index out of range

ImistroI commented 5 months ago

I have the same issue, did you fix it?

ImistroI commented 5 months ago

FYI: I fixed the problem. my ffmpeg was corrupted. Whenever I tried to call ffmpeg from the terminal I got this message: "ffprobe: error while loading shared libraries: libopenh264.so.5: cannot open shared object file: No such file or directory" I have solved it by downloading from here my binary: https://github.com/cisco/openh264/releases/tag/v2.1.1 then extract it and renaming it to libopenh264.so.5 and then moving it to "${CONDA_PREFIX}/lib". Also I add in the emoca/gdl/utils/FaceVideoDataModule.py the import ffmpeg (not sure if needed but on my system works). Also the test video from there didn't work but my own input video works (don't know why)

swerizwan commented 5 months ago

Yeah, I have also fixed the issue, in my case I think there was the wrong path to the video.

zizimars commented 4 months ago

I faced the same issue. But after running on my own video it finally works

wang5d commented 3 months ago

I faced the same issue，I changed my own video and still reported this error，Do you know how to fix this