Datasets:
Request sample access
This repository is publicly accessible, but you have to accept the conditions to access its files and content.
This listing is a preview. The production dataset (100,000 hours of 1080p+ first-person residential task video with structured annotations) is rights-cleared and delivered directly under a commercial license. Approved requesters get the full annotation schema and data dictionary in this repository, and can request a review package with real video samples and QA summaries. Requests are reviewed within 1 business day.
Log in or Sign Up to review the conditions and access this dataset content.
Egocentric Residential Video Dataset
100,000 hours of 1080p+ first-person video of real household chores recorded in homes across Brazil, with structured annotations for tasks, actions, objects, hand-object interactions, and completion states.
This repository is a specification and preview listing. The production dataset is rights-cleared and delivered directly to buyers. Request access to see the full schema and get real video samples.
Overview
The Egocentric Residential Task Video Dataset is a 100,000-hour collection of high-quality first-person videos recorded in real residential environments across Brazil at 1080p resolution or higher. It captures natural household chores from the human point of view, preserving hand-object interactions, room context, object placement, surfaces, tools, and multi-step task flows.
Built for Physical AI, robotics, egocentric vision, household automation, manipulation learning, and embodied AI, the dataset focuses on non-cooking residential tasks: cleaning, laundry, organizing, bathroom maintenance, dish handling, pet-area upkeep, plant care, and basic home maintenance.
Annotations capture not only what action is happening, but how it fits into a larger residential workflow: task phases, pickup and placement events, and before-and-after household state changes.
At a glance
| Total volume | 100,000 hours |
| Perspective | Egocentric (first-person) |
| Resolution | 1080p or higher |
| Frame rate | Commonly 30fps, higher-frame-rate subsets where available |
| Geography | Residential environments across Brazil |
| Task types | 80 non-cooking household chore categories |
| Environments | Bedrooms, bathrooms, kitchens, laundry areas, living rooms, closets, storage, balconies, patios, utility areas |
Task coverage
80 non-cooking household task types, including:
making beds, changing bedsheets, replacing pillowcases, folding blankets, sorting laundry, loading and unloading washing machines, hanging clothes, folding garments, organizing drawers and closets, collecting trash, replacing trash bags, sorting recycling, sweeping, vacuuming, mopping, dusting, wiping tables and counters, cleaning mirrors and windows, sanitizing door handles, organizing living rooms, unpacking packages, cleaning bathroom sinks, toilets, showers, and bathtubs, restocking supplies, washing dishes by hand, loading and unloading dishwashers, cleaning pet areas, refilling pet water bowls, watering plants, sweeping patios or balconies, and replacing light bulbs.
Technical specifications
- Annotation coverage: task type, action segments, room labels, object labels, tool labels, surface labels, hand-object interactions, task phases, completion states, before/after state changes
- Interaction metrics: pickup events, placement events, wiping/scrubbing events, open/close events, object transfer events
- QA metrics: video usability score, blur/exposure checks, hand visibility percentage, annotation quality score, completion label review
- Privacy: PII redaction, face blur where needed, address/identifier removal, LGPD-aware review
- Delivery format: MP4 video, CSV metadata, JSON annotations, thumbnails or preview frames
Hand visibility is tracked per clip because it is the single strongest predictor of downstream manipulation-learning value; low-visibility footage is filtered at QA.
Annotation schema
The gated file annotation_schema.json in this repository contains the full clip schema with an illustrative example record, including action segments, hand-object interaction events, and before/after state annotations.
How to evaluate this dataset
- Request access using the form above. Requests are reviewed within 1 business day.
- On approval you get the gated files in this repository: full annotation schema, data dictionary, and access instructions.
- Request a review package and we deliver real video samples in your target task categories, JSON annotations, metadata CSVs, QA summaries, and licensing documentation within 2 business days.
All samples are delivered with structured CSV metadata and JSON annotation files where available. Buyer review packages include representative media files, metadata samples, annotation schema, QA summaries, and data dictionary documentation.
Licensing
The production dataset is rights-cleared for commercial AI training and licensed directly by Datoric, with documented contributor consent, LGPD-aware privacy review, and chain-of-custody. Subset (by task category or hours), exclusive, and custom-collection options are available.
About Datoric
Datoric supplies rights-cleared, spec-exact training data for frontier AI labs and enterprise model teams: egocentric and industrial video, human manipulation data, robot episodes, expressive multilingual voice, computer-use traces, and gameplay trajectories. We also run managed collection pipelines for custom specifications.
Contact: nikhil@arzule.com
- Downloads last month
- 25