search

rohitgirdhar / CATER

CATER: A diagnostic dataset for Compositional Actions and TEmporal Reasoning
https://rohitgirdhar.github.io/CATER/
Apache License 2.0
103 stars 19 forks source link

corrupted videos in pre-generated data #27

Closed vmichals closed 3 years ago

vmichals commented 3 years ago

There are several corrupted video files in the downloadable zip datasets, e.g. 'videos/CATER_new_004798.avi' in the max2action task (see screenshot). We downloaded the archive multiple times and the same files are corrupted. How should we handle this?

Code to see the issue:

wget https://cmu.box.com/shared/static/jgbch9enrcfvxtwkrqsdbitwvuwnopl0.zip
# extract one corrupted video
unzip videos.zip videos/CATER_new_004798.avi
# open in a media player:
# vlc videos/CATER_new_004798.avi
# mplayer videos/CATER_new_004798.avi
rohitgirdhar commented 3 years ago

Hi, thanks for your interest. Yes unfortunately there are some corrupted videos as except for some basic automated checks here, we were unable to manually verify that all the videos were rendered/stored correctly. You can choose to ignore those videos, I believe our baseline code either ignores or returns a dummy/mean-image like output for any video it can not decode.

vmichals commented 3 years ago

Thanks for the quick reply! We'll simply skip them for now!