About our egocentric video annotation
Raw first-person video becomes training data once it is labelled. Our annotators mark when each action starts and ends, which objects are used, where the hands are, and what is happening in plain language.
Where it's used
- Action recognition
- Video-language models
- Robot learning from video
- Benchmarks
What we offer
Action segmentation
Start and end times for every step, from your list of actions.
Objects and hands
Boxes, masks and tracks for hands and the objects they touch.
Narrations and captions
Short, timestamped descriptions of what the person is doing.
Quality checks
Consensus between annotators and reviewer audits, as on every project.
From first call to final delivery
Scope
We agree the data, the quality bar, the formats and a sample size, and write it into clear guidelines.
Pilot
A small batch first. You review it, we tune the guidelines, and only then do we scale.
Produce
Vetted contributors work in our studios. Automatic checks catch problems as files arrive.
Verify
Consensus between annotators, hidden check items and reviewer audits on every batch.
Deliver
Clean exports in your format, with documentation and a quality report.
What you receive
- MP4 video (1080p or more, 30 fps or more)
- Per-clip metadata: scenario, device, lens, length, sensor data
- Consent and site-permission records
- Quality report with automatic and reviewer checks
- Optional labels: action steps, hands, objects, narrations