EgoTracks Dataset
Long-term object tracking in egocentric video
An Ego4D benchmark focused on tracking objects through heavy hand interaction, occlusion, viewpoint changes, and object disappearance/reappearance.
Object persistence in robot tasks
Useful for Bench/Data integration when failures involve losing an object through a manipulation step.
Inspect schema and run a bounded sample audit before committing to the full release.
Has observation, action/state proxy, and task or language context.
fit 85 · confidence 55
Has rich observation and semantic context, but limited geometry/sim-real alignment.
fit 68 · confidence 55
Contains observation, intent, action/state, and feedback-like supervision.
fit 88 · confidence 55
Verified facts and provenance
Claims, metadata verification, and sample verification are shown separately.
Unknown — no machine-readable schema facts have been captured.
Unknown — metadata conclusions do not prove sample coverage, alignment, or file integrity.
Declared loop signal coverage
Signals inferred from official metadata; Data pipeline verification is still pending.
Observation / ego video
Ego4D video · Long-term object tracking in egocentric video
Action / hand pose / robot state
Occlusion-heavy hand-object tracking · Useful for Bench/Data integration when failures involve losing an object through a manipulation step. · An Ego4D benchmark focused on tracking objects through heavy hand interaction, occlusion, viewpoint changes, and object disappearance/reappearance.
Gaze / attention
No decision-grade evidence captured yet.
Language intent / task phase
JSON annotations · Object persistence in robot tasks
Feedback / correction / failure
Occlusion-heavy hand-object tracking · Failure replay around lost targets · Useful for Bench/Data integration when failures involve losing an object through a manipulation step. · Long-term object tracking in egocentric video
Sim-real pairing
No decision-grade evidence captured yet.
License / format / access
License required · Ego4D License Agreement · JSON annotations · Ego4D video
Model and task fit · OpenBot inference
Has observation, action/state proxy, and task or language context.
fit 85 · confidence 55
Has rich observation and semantic context, but limited geometry/sim-real alignment.
fit 68 · confidence 55
Contains observation, intent, action/state, and feedback-like supervision.
fit 88 · confidence 55
Has failure/evaluation-style labels with action or manipulation context.
fit 82 · confidence 55
Good tasks
Blockers and unresolved evidence
- Gaze / attentionunknownNot enough evidence to classify this signal. Verify metadata or a bounded sample.
- Sim-real pairingunknownNot enough evidence to classify this signal. Verify metadata or a bounded sample.
Raw dataset signals
OpenBot fit
- Object persistence in robot tasks
- Occlusion-heavy hand-object tracking
- Failure replay around lost targets
Related models and papers
Model references linked to similar loop signals.
OA-WAM
Object-addressable world-action model that decomposes scenes into robot and object slots while jointly predicting future world state and actions.
LingBot-Vision
A family of vision foundation encoders pretrained for dense spatial perception and downstream embodied understanding.
Integration notes
- Useful for Bench/Data integration when failures involve losing an object through a manipulation step.
- Part of Ego4D, so access follows Ego4D terms.