VIDEOEGOCENTRIC
First-Person Activity Clips
Egocentric activity video for models that need to understand actions, objects and task progression from the actor's point of view.
View datasetVIDEOPROCEDURES
Procedural Demonstration Video Sets
A procedure is defined by order as much as content. Procedural Demonstration Video Sets turn multi-step activities into temporally structured data so a model can work with the sequence of actions rather than only the final video category.
Steps can be localised on the timeline and connected to progress or completion information. When narration is part of the demonstration, it can also be aligned to the relevant stage of the procedure, giving models another signal for how an action is described and performed.
Technical structure
- Ordered procedure steps with temporal boundaries
- Progress and completion information linked to the sequence
- Optional narration and transcript alignment where useful
- Project-defined procedure or step identifiers
Typical applications
Step segmentation, next-step prediction, procedural understanding, progress estimation and completion detection.
Delivery note
Step definitions and granularity are set for the project rather than assumed from a universal procedure taxonomy.
Delivery
The exact package depends on the collection scope. Where relevant, delivery can include task-specific records, manifests, provenance fields, stable identifiers, SHA-256 hashes, MLCommons Croissant 1.0 metadata and loading instructions.