Annotation services

Video annotation & object tracking services

Video annotation labels objects across the frames of a clip and gives each one a persistent track ID, so the same car or person keeps one identity from entry to exit. Annotators label keyframes, interpolate between them and correct drift, occlusion and re-entry, producing training data for detection and multi-object tracking models.

Trains
Object tracking and video detection
From
$0.057 / label
Review
Every job
Formats
MOT challenge TXT

Is it the right label?

When to use video & object tracking

Object tracking and video detection — the model follows each object through time, not frame by frame in isolation.

  • Identity over time matters: counting unique people or vehicles, dwell time, trajectories, re-identification.

  • Your model runs on video and needs temporally consistent labels.

  • You need action or event boundaries — when something starts and stops.

When something else fits better

If you only need detection and frames are far apart, sampling stills and labeling them as images is cheaper. We can advise on a sampling rate during scoping.

Bounding boxes

How we label it

Our process

  1. 01

    Sample the right frames

    Frame rate and sampling are agreed up front — labeling every frame of 30 fps footage is rarely needed.

  2. 02

    Keyframes and interpolation

    Annotators place keyframes where motion changes and interpolate between them, then correct every frame where interpolation drifts.

  3. 03

    Identity through occlusion

    When an object is hidden and reappears it keeps its original ID. Leaving and re-entering the frame follows the guideline's rule.

  4. 04

    Track review

    Reviewers scrub each track for ID switches, drift and missing segments before the clip is approved.

Quality

What our QA checks

Every job is reviewed by a person other than the annotator, on top of automated checks. Work below the agreed threshold goes back for rework.

ID-switch detection across each track

Per-frame box IoU on sampled frames

Track continuity — no gaps where the object is visible

Start/end frame accuracy on events

Pricing & delivery

Rates and formats

AnnotationRate
Object detection$0.057 / label
Instance segmentation$0.14 / label

QA review is included. Bulk discounts apply to large projects; the final quote is confirmed after we review a sample of your data. Full pricing

Delivered as

MOT challenge TXTCOCO JSON (per frame)YOLO TXT (per frame)CVAT video XMLDatumaro

Used for

Autonomous driving and ADAS · Retail footfall and dwell time · Sports player tracking · Surveillance and safety · Traffic analytics

Video & object tracking FAQ

Video & object tracking, answered

What is video annotation?

Video annotation is labeling objects across the frames of a video, usually with a persistent ID per object so it can be tracked over time. It trains video detection and object tracking models.

Do you annotate every frame?

Usually not. We agree a sampling rate with you, label keyframes, interpolate between them and correct any frame where the interpolation drifts. The result is a label on every delivered frame at a fraction of the effort.

How do you keep object IDs consistent?

Each object gets one track ID from first to last appearance. Annotators carry the ID through occlusions, and reviewers scrub every track for ID switches before delivery.

How is video annotation priced?

Per labeled object per delivered frame, so the frame sampling rate and the number of tracked objects decide the total. We quote after seeing a sample clip.

Which video annotation formats do you deliver?

MOT challenge format, CVAT video XML, Datumaro, or per-frame COCO and YOLO files with track IDs — whichever your tracking pipeline reads.

Need video & object tracking on your data?

Send us a few sample images or clips. We'll agree the guideline with you and send a plan and a per-label quote.