Missing videos when downloading EgoExo4D dataset

Hello,

I’m building the TokenBench dataset, which includes 200 videos from the EgoExo4D project.

I’ve been using the Ego4D CLI to download both v1 and v2 videos.
However, I’ve encountered an issue where several videos are unavailable in my downloaded dataset.

The following videos are missing:

/mnt/cnsmldb108/user/egoexo4d/takes/cmu_bike01_3/frame_aligned_videos/aria01_214-1.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/cmu_bike01_3/frame_aligned_videos/cam01.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/cmu_bike15_1/frame_aligned_videos/aria01_214-1.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/cmu_bike15_1/frame_aligned_videos/cam01.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/cmu_soccer14_3/frame_aligned_videos/aria01_214-1.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/cmu_soccer14_3/frame_aligned_videos/cam01.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/georgiatech_covid_06_11/frame_aligned_videos/aria01_214-1.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/georgiatech_covid_06_11/frame_aligned_videos/cam01.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/iiith_cooking_100_6/frame_aligned_videos/aria01_214-1.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/iiith_cooking_100_6/frame_aligned_videos/cam01.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/indiana_music_01_5/frame_aligned_videos/aria01_214-1.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/indiana_music_01_5/frame_aligned_videos/cam01.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/nus_cpr_26_4/frame_aligned_videos/aria01_214-1.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/nus_cpr_26_4/frame_aligned_videos/cam01.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/sfu_cooking017_8/frame_aligned_videos/aria01_214-1.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/sfu_cooking017_8/frame_aligned_videos/cam01.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/sfu_cooking022_4/frame_aligned_videos/aria01_214-1.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/sfu_cooking022_4/frame_aligned_videos/cam01.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/sfu_cooking022_8/frame_aligned_videos/aria01_214-1.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/sfu_cooking022_8/frame_aligned_videos/cam01.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/unc_soccer_09-21-23_01_21/frame_aligned_videos/aria01_214-1.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/unc_soccer_09-21-23_01_21/frame_aligned_videos/cam01.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/unc_soccer_09-22-23_01_26/frame_aligned_videos/aria01_214-1.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/unc_soccer_09-22-23_01_26/frame_aligned_videos/cam01.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/unc_soccer_09-22-23_02_14/frame_aligned_videos/aria01_214-1.mp4
/mnt/cnsmldb108/user/egoexo4d/takes/unc_soccer_09-22-23_02_14/frame_aligned_videos/cam01.mp4

I’ve checked the Ego4D dataset viewer, and it appears that at least some of these videos (e.g., cmu_bike01_3) exist in the dataset.

Could you please provide guidance on how to download these missing videos? Are they part of a different subset, or do they require a separate download process?

Your assistance in resolving this issue would be greatly appreciated, as it’s crucial for completing the TokenBench dataset.

Thank you for your time and support.

Best regards,
Donghyun Kim

We removed some videos from V1 → V2 due to data quality.

All takes you have referenced were in-between takes (no action is being performed, camera setup, testing shot, etc.). To give you full context, the universities reviewed them and they marked them to not be included due to them being in-between takes or for privacy concerns.