VIDEO & EGOCENTRIC DATA
Capture whatthe model needs.
First-person and task-specific video, scoped around what the model needs to see.
Start from the task.
Define the actions, viewpoints, settings, contributors, permissions and formats before collection.
From brief to delivery.
01
Scope
Task definitions, locations, contributors, rights and formats.
02
Capture
Collection instructions, coordination and review.
03
Enrich
Optional temporal, spatial and task-specific labels.
04
Deliver
Approved footage, metadata and agreed documentation.
Collection formats.
Options depend on the task and capture plan:
- 01Egocentric video
- 02Human-object interactions
- 03Procedural tasks
- 04Multi-scene footage
- 05Task-segmented clips
Plan a video dataset.
Tell us what the model must see. We’ll scope the capture plan.