# Assembly101

From the catalogue. Listed by one researcher from one source page.

- id: `assembly101`
- kind: egocentric
- robot or device: Headset with 4 egocentric cameras plus 8 static cameras (class: human egocentric)
- how collected: participants assemble and disassemble 101 toy vehicles in a camera cage
- size as stated: 4,321 videos; 513 hours (per awesome list)
- hours: 513 | hours only claimed: none | episodes: not stated
- year: 2022
- organisation: Meta Reality Labs and NUS
- licence: not stated (class: not stated)
- access: open (Hugging Face since May 2026) (class: open)
- link (dataset): https://assembly-101.github.io/
- note: 1M fine action segments, 18M 3D hand poses.

Confirm the licence at the link before relying on it.