What the hands see, at the moment they see it.
Monocular first-person video from a chest or head mount, recorded by people doing everyday tasks in the rooms where those tasks actually happen — for imitation learning, hand–object understanding and embodied pretraining.
- Environment
- Apartment kitchen, mid-morning window light
- Captured with
- GoPro on a chest mount, RGB + on-device audio
- Task
- Empty a cupboard shelf, wipe it, put the containers back
Perit has shipped 4,000+ hours of speech through this bench. Physical-AI capture runs on the same recruiting, consent and QA machinery, and the first collections are being scoped with early-access partners now. Nothing on these pages is a delivered volume — the first hours will be listed here the way our speech hours are, after they ship.
What gets captured, and why it is worth having.
A camera on the chest or forehead, a task, and a room. The person is not an actor: they are cleaning a stove, folding clothes, restocking a shelf, and the footage carries everything a staged capture loses — hesitation, reach, re-grasp, the way a cupboard door actually swings.
Because the rig is a phone or an action camera, the pool is anyone who can follow a protocol. That is what makes egocentric the fastest way to put thousands of diverse hours in front of a model before a single robot exists.
- Imitation learning from human video
- Hand–object interaction and affordance models
- Task segmentation and long-horizon activity understanding
- Embodied pretraining ahead of robot data
- Primary rig
- GoPro-class action camera on a chest or head mount
- Also
- iPhone or Android with a timestamped IMU stream
- Also
- Custom rigs to your spec — stereo, wider field of view
- Video
- H.264 in MP4
- Resolution
- 4K or 1080p
- Frame rate
- 29.97 / 30 fps
- Audio
- Mono, on-device
- IMU
- Accelerometer and gyroscope, timestamped to frames on phone rigs
- Episode
- 2–15 minutes, one task each
Where it is recorded.
What people are asked to do.
Tasks are recorded where they actually happen — not in a studio dressed to look like a kitchen. Hover a room to see the kind of setting the protocol calls for.
- Raw video per episode, original device capture kept
- Task and environment metadata per episode
- QC status and reviewer chain per episode
- Consent reference on every file
- Quality report per batch: accepted, rejected, and why
- Weekly drops to your bucket, region pinned at kickoff
- Task and sub-task segmentation
- Natural-language captions
- 3D hand pose estimation
- Object and interaction labels
Labelled on the same bench as our speech work, to a written guide, behind the same calibration test.
Small first, on purpose.
- 01Briefday 0
The task family, the environments, the rig, the sensors, the episode count. One page, agreed before anything is bought or anyone is recruited.
- 02SOP and kitweek 1
A written capture protocol — framing, lighting, where a task starts and ends, what counts as a failed take — and the kit list. Operators are trained and tested on it before a single episode is paid for.
- 03Pilot episodesweek 2
A small batch, ten to thirty episodes, delivered with QC status and a quality report. This is where the protocol breaks — on purpose, and cheaply.
- 04Review and scaleongoing
You review the pilot, we fix the protocol, then the run scales on a weekly delivery cadence with the same report attached to every batch.
Scope a egocentric pilot.
The task family, the environment, the rig and the episode count — one page, and we come back with a protocol and a quote.