Home / Services / Data labelling

Video annotation

Objects tracked across frames with stable IDs, and actions marked with exact start and end times, including first-person video.

Dashcam clip, frames 120 to 150QA passed
car ID 07 person ID 12f120f130f140f150
car ID 07person ID 12
AnnotatorLabels drawn
ReviewerChecked
QAApproved

What you get

  • Object tracking with persistent IDs
  • Action and activity segments with timestamps
  • Hand-object interaction tags for egocentric clips
  • Keyframe or every-frame labelling
  • Clip-level summaries and captions

How we check quality

  • ID switches checked through each clip
  • Interpolated frames reviewed by a person
  • Segment boundaries checked frame by frame
  • Clip-level samples re-labelled by QA

How it works

A clear path from first conversation to delivered data.

  1. GuidelinesWe agree classes, actions and frame rules.
  2. PilotA few clips are labelled free for your review.
  3. LabelAnnotators track and tag clip by clip.
  4. Review and deliverReviewers watch every clip in full before QA.

Common questions

Do you label every frame?

Either every frame or keyframes with checked interpolation, depending on your needs and budget.

Can you label long videos?

Yes. We split long videos into clips with overlap so IDs stay consistent.

Which formats do you deliver?

MOT, CVAT video XML, COCO video JSON, CSV or JSON timestamps.

Start with a free pilot

Send a small sample and your guidelines. No cost, no commitment.

Request a free pilot