Privacy and Consent for Egocentric Data
Operational controls for first-person recording in homes, workplaces, and other real environments.
Read Guide ↗Quality and governance
Build Quality, Rights, and Provenance Into the Data Program.
Connect capture quality, privacy, licensing, lineage, security requirements, and delivery checks from the first protocol to the final files.
Buyer requirements
Set Quality and Governance Requirements Before Collection.
Define privacy, rights, security, retention, documentation, and delivery requirements alongside the data specification.
What Exists Today, and What You Can Specify.
Separate the data you can review now from the details that are scoped for a specific collection or annotation program.
- Available to licenseHousehold egocentric video dataset
- What EGXO provides
License from EGXO’s continuously growing first-party household egocentric catalogue. Current coverage, volume, delivery format, and rights are reviewed against buyer requirements. This inventory was collected through licensed GIG Rewards collection programs operated with telco partners.
- Set for your project
Confirm dataset fit, intended use, pricing, permitted rights, security, retention, delivery, and any required enrichment in writing.
- Custom programCustom egocentric program
- What EGXO provides
Design targeted collection around the behavior, setting, viewpoint, and signal your model needs to learn.
- Set for your project
Define contributor eligibility, task, environment, viewpoint, volume, turnaround, rights, and acceptance criteria.
- Custom programExocentric capture
- What EGXO provides
Design fixed or multi-angle capture around whole-body motion, workspace geometry, manipulation, navigation, and evaluation.
- Set for your project
Set camera topology, coverage, occlusion limits, calibration, rights, and ingest requirements in the pilot.
- Custom programSynchronized ego/exo capture
- What EGXO provides
Pair actor-aligned and external views for cross-view learning, pose, occlusion recovery, and independent evaluation.
- Set for your project
Specify the timebase, drift tolerance, calibration error, view association, and loader validation criteria.
- Custom programData annotation and enrichment
- What EGXO provides
Turn existing video, image, or sensor data into model-ready supervision with a project-specific ontology, workflow, review method, and delivery schema.
- Set for your project
Define the learning objective, label types, granularity, boundary rules, acceptance thresholds, adjudication, lineage, and export format.
- Project-specificSensors and metadata
- What EGXO provides
Add RGB, metadata, sensor, language, pose, or other signal requirements around the model objective.
- Set for your project
Select the required signals, units, synchronization, provenance, QA method, and end-to-end ingest test.
- Project-specificDelivery format and license
- What EGXO provides
Define the delivery around the buyer's ingest path, documentation needs, permitted use, and governance requirements.
- Set for your project
Confirm schema, loader, documentation, retention, derived-model rights, and redistribution terms in writing.
Share the use case and we will identify whether current inventory fits or define the collection, annotation, deliverables, success criteria, rights, and project plan.
Discuss Data Options01
Plan
Version the task, capture setup, environment, devices, instructions, and stop rules.
02
Capture
Track media properties, timestamps, checksums, required fields, and asset relationships.
03
Review
Check visibility, task completion, annotations, privacy, rights, rejection, and adjudication.
04
Deliver
Package the dataset card, schema, version, lineage, license, limitations, and ingest results.
Quality Checks Should Test What the Model Actually Needs
Validate the files, task visibility, labels, coverage, rights, and delivery—not just whether a video plays.
Media
Integrity
Decode, duration, dimensions, codec, frame rate, audio policy, corruption, and duplicates.
Task
Observability
Required actions, contacts, objects, outcomes, and failures remain visible.
Labels
Consistency
Ontology, boundaries, agreement, confidence, adjudication, and version.
Context
Coverage
Task, environment, object, viewpoint, and edge-case distributions match the brief.
Rights
Eligibility
Consent, privacy review, permitted use, distribution tier, and retention status.
Delivery
Reproducibility
Schema, checksums, splits, transformations, lineage, and loader validation.
Provenance
Keep Every Asset and Transformation Traceable.
Raw assets, normalized media, clips, annotations, redactions, exports, and releases should use stable identifiers and versioned lineage. Corrections must be additive and auditable rather than silently overwriting the record.
Buyer-facing documentation should explain what changed, which tools or reviewer roles performed the change, and which source and policy versions produced the delivered output.
Read the Dataset Provenance Definition ↗Apply These Controls to Data Annotation ↗Set Rights and Privacy for the Intended Use
Collection, annotation, model training, buyer delivery, retention, and public presentation require clearly defined permissions.
- 01
Minimize
Exclude unnecessary people, places, screens, documents, audio, and sensitive context before recording.
- 02
Document
Connect consent, rights scope, policy version, compensation, and retention to eligible assets.
- 03
Screen
Review for incidental identity, private information, sensitive activity, and distribution restrictions.
- 04
Control
Apply access tiers, least privilege, revocable credentials, logging, and contractual limits.
- 05
Reapprove
Run the release gate again for new versions, buyers, public pages, or distribution channels.
Project requirements
Put Quality and Rights Into the Project Brief.
Share the privacy, rights, security, quality, documentation, and delivery requirements for your program.
Discuss Requirements