Human demonstrations · Meta
HOT3D
Egocentric multi-view hand and object tracking with 3D ground truth.
Release / update
Object-onboarding sequence release update, including Aria sequences.
What is included?
- Data type
- Human demonstrations
- Tasks
- Hand-object interaction · Multi-view perception
- Modalities
- RGB · Monochrome video · Hand pose · Object pose
- Embodiment
- Human participants wearing Aria or Quest 3
Reported scale
833 recording minutes · 1.5M multi-view framesPublisher-reported scope. Preserve the stated unit; no duration or trajectory conversion is inferred.
Access & license
Follow the official full-dataset or HOT3D-Clips instructions and the HOT3D license agreement.
Before you use this dataset
Full recordings and curated clips are not interchangeable. Multi-view frames and per-camera images use different counts.
Publisher sources checked 2026-10-05. Listing review checks the documentation, not downloaded data quality. Always check the exact release and terms.