Some discussion of this here: https://github.com/cleanlab/cleanvision/discussions/214 Making CleanVision runnable on video would be super useful to many folks! Should be fairly easy: - just extract every k-th frame from the video, - run cleanvision on all those images, - aggregate the results across frames per video