DatoricEgocentric Residential Video Dataset
Available · 80 task categories
Train physical AI and robotics systems on first-person household demonstrations that preserve hands, objects, room context, task phases, and visible state changes.
Dataset specification
- 1080p+ first-person video
- 80 non-cooking residential task categories
- Task, action, object, and room labels
- Hand-object interaction labels
- Before/after state changes
- Hand visibility metrics
- QA review
- Contributor consent documentation
- Privacy-redaction review
Intended uses
- Physical AI and embodied-agent training
- Household robotics and manipulation learning
- Egocentric action and hand-object interaction modeling
- Task planning and visible state-change prediction
Scope notes
- Task, environment, contributor, and collection-window distributions are confirmed for the selected delivery.
- Frame rate and higher-frame-rate coverage vary by subset and are documented during review.
- Repository metadata illustrates the schema; real video is supplied in the buyer review package.
Collection scope
Environments
- Real residential environments
- Bedrooms, bathrooms, kitchens, laundry, living, storage, and utility areas
- 1080p or higher first-person capture
Task coverage
- 80 non-cooking residential task categories
- Cleaning, laundry, organizing, dish handling, pet-area upkeep, plant care, and home maintenance
- Multi-step household workflows
Available scale
- Available program
- First-person video with structured task and interaction metadata
Buyer review package
- Annotation schema, data dictionary, and sample metadata in the repository
- Real video, JSON annotations, metadata, and QA summaries for review
- Target tasks, environments, contributors, capture window, and hours confirmed for the selected delivery
Annotation & metadata fields
- Task type and action segments
- Room, object, tool, and surface labels
- Hand-object interaction events
- Pickup, placement, wiping, scrubbing, open, close, and transfer events
- Task phases and completion states
- Before and after state changes
- Hand visibility and video usability metrics
- Privacy-redaction review fields
Capture methodology
- 1080p or higher egocentric video recorded from the contributor's point of view
- Household tasks captured as complete, multi-step workflows in residential environments
- Action segments aligned with room, object, tool, surface, and interaction labels
- Before and after states recorded to preserve visible task outcomes
Provenance & rights chain
- Contributor consent documentation included in dataset review
- Commercial license and chain-of-custody record issued directly by Datoric
- Privacy-redaction requirements recorded with the selected delivery
- Collection and annotation activity handled under Datoric's published privacy notice
Quality, duplicates & PII
How submissions are reviewed and cleaned before they are accepted into the dataset.
- Video usability, blur, exposure, and hand-visibility checks
- Annotation quality and completion-label review
- PII redaction, face blurring where needed, and identifier removal
- Duplicate, malformed, and incomplete records handled under the agreed acceptance criteria
Formats & delivery
- MP4 video
- CSV metadata
- JSON annotations
- Thumbnails or preview frames
Rights & license scope
Licensed directly by Datoric for commercial AI training, with final scope controlled by the signed agreement for the selected delivery.
Version & verification
- Availability
- Available
- Datasheet version
- July 21, 2026
- Release date
- July 21, 2026
- Last verified
- July 21, 2026
- Owner
- Datoric
Frequently asked
- Can we review real household video?
- Yes. The buyer review package includes real video for selected task categories, JSON annotations, metadata, QA summaries, and licensing documentation.
- Which household tasks are covered?
- The program spans 80 non-cooking categories across cleaning, laundry, organizing, dish handling, pet-area upkeep, plant care, and basic home maintenance.
- How is privacy handled in residential footage?
- The specification includes PII redaction, face blurring where needed, identifier removal, and privacy-review fields that can be inspected with the delivery documentation.
- Can we target a narrower environment or task set?
- Yes. Review and pilot scope can be selected by task category, environment, annotation needs, and hours before production delivery.