libav.h264 ERROR for AV-Deepfake1M++

[zlin0]

Hi,

when I following read_video() to deal with mp4 files from AV-Deepfake1M++:

def read_video(path: str):
    video, audio, info = torchvision.io.read_video(path, pts_unit="sec")
    video = video.permute(0, 3, 1, 2) / 255
    audio = audio.permute(1, 0)
    if audio.shape[0] == 0:
        audio = torch.zeros(1, 1)
    return video, audio, info

I got some libav.h264 ERROR as below:

ERROR:libav.h264:Invalid NAL unit size (2017003656 > 5787).
ERROR:libav.h264:missing picture in access unit with size 5791
ERROR:libav.h264:Invalid NAL unit size (187024585 > 4974).
ERROR:libav.h264:missing picture in access unit with size 4978
ERROR:libav.h264:Invalid NAL unit size (572622029 > 5104).
ERROR:libav.h264:missing picture in access unit with size 5108

enviorment: torchvision: 0.22.1
Has anyone encountered the same issue? Did I download broken files?

Thank you!