search

EGO4D / episodic-memory

MIT License
108 stars 60 forks source link

Clips in VQ2D annotation seem to be different from benchmark clips. #17

Closed hskAlena closed 2 years ago

hskAlena commented 2 years ago

Hello! I've checked val_annot.json.gz in vq2d and I found out 'video_frame_number' and '(clip)frame_number' point to other frames, shown in the first image. But when I used extracted clips from full_scale videos (using convert_videos_to_clips.py), not benchmark clips I downloaded directly from Ego4D, it seems to return frames within the same range, as in the second image.

Is this right? The benchmark clips are not for VQ2D annotations?

image

image

srama2512 commented 2 years ago

@hskAlena - The benchmark clips are likely at a different frame rate than is required for VQ2D (which needs 5 FPS). The convert_videos_to_clips.py accounts for the frame rate changes. You could modify the script to process clips instead of videos to get the VQ2D 5 FPS clips.

srama2512 commented 2 years ago

@hskAlena - can we close this issue? Did the above response address your question?

hskAlena commented 2 years ago

Yes! The frame rates are different. Thanks for the answer!