Thanks for helping build the definitive data quality catalog for AI/ML datasets.
Open a Dataset Submission issue. We'll verify and add it.
Open a Data Correction issue with the current value, correct value, and your source.
Open a Feature Request issue.
Fork, make changes, and submit a PR using our template. All PRs must pass the verification checklist.
Category — must be one of: Large-scale foundation, Classic, Action recognition, Hand-object interaction, Body pose, Gaze and eye tracking, Industrial and procedural, Social interaction, Ego-exo cross-view, Navigation and localization, 3D reconstruction, Object detection, VQA and language, Audio-visual, Driving, Video generation, Lifelogging, Surgical, Child egocentric, Drone FPV, Sports, Expert assistance, Assistive, Animal, Tracking
Access Level — must be one of: Open, Gated-Open, Restricted, Commercial, Unavailable
License Type — must be one of: CC-BY-4.0, CC-BY-NC-4.0, CC-BY-SA-4.0, CC-BY-NC-SA-4.0, CC-BY-NC-ND-4.0, MIT, Apache 2.0, Custom Academic, Custom Open, S-Lab License, Mixed, Not specified
URL Status — must be one of: Live, Live but broken, Dead, Redirect
Modalities — semicolon-separated, using these terms: RGB video, Depth (RGB-D), Depth (stereo), Audio, IMU, Eye gaze, Hand pose annotations, Body pose annotations, Force/torque, Tactile, Motion capture, SLAM/odometry, 3D point clouds, 3D scene reconstruction, LiDAR, Event camera, Optical flow, Narrations/captions
- Total Hours: plain numbers only. No parentheticals. "~10" not "~10 (43 subjects x 5 recipes)".
- Number of Participants: plain numbers only. "37" not "37 participants; 45 kitchens".
- No emoji in any CSV cell.
- License check: before marking "Not specified", check website, GitHub LICENSE file, paper appendix, and download page. Note what you checked in Key Findings/Issues.
- Paper over website: when info conflicts, trust the paper.
Every dataset entry must have:
- Primary URL actually fetched (not guessed)
- Download mechanism verified (not just landing page)
- License checked across 4 sources
- All fields using standardized vocabulary