First-person video · Embodied AI
Egocentric data.
A human perspective.
First-person video puts a model close to the work: hands, objects, actions, and the environment around them. Tell us which perspective your research needs.
Discuss your requirements Read the sourcing guidesCategory expertise. A direct line to Fiund.
The category, in context
An egocentric dataset records an activity from the participant’s point of view. A useful sourcing brief goes beyond hours of footage: it defines the task, camera position, visible interactions, and conditions the model must learn to handle. Data Egocentric is a Fiund sourcing site for teams exploring this category.
01 / Sourcing focus
Define the collection
around the use case.
Everyday activities
Preparation, assembly, tidying, and other task sequences. Describe the actions and environments you want represented, including the moments between major steps.
Hands and object interaction
Grasping, tool use, handoffs, and manipulation viewed from the participant. Specify whether hands, contacted objects, and task outcomes must remain visible.
Workplace demonstrations
First-person recordings of practical processes. Identify permitted environments, participant requirements, sensitive areas, and the kind of documentation your project needs.
02 / Before you source
The details that
make a brief useful.
Viewpoint
Head-mounted, chest-mounted, or another participant-centered view; describe acceptable camera movement.
Task coverage
Action list, sequence length, successful and unsuccessful attempts, and variation between participants.
Capture details
Resolution, frame rate, field of view, audio policy, and any requirements for time alignment.
Annotation
Action boundaries, object references, task descriptions, and the level of detail needed for evaluation.
These are categories and specification considerations for inquiry, not a live inventory listing. Specific collections and permitted uses require review.
03 / Field guides
A better brief starts
with better questions.
How to scope an egocentric data collection
A practical brief for first-person tasks, camera placement, coverage, and review.
An egocentric video dataset evaluation checklist
Review viewpoint, task completeness, metadata, and permitted use before selecting first-person data.
04 / Common questions
A few things
to clarify.
Is egocentric video the same as robot demonstration data?
No. A participant’s camera can show a human performing a task without recording robot actions or control signals. Tell us if you need synchronized robot state, action data, or other sensors; video alone should not be treated as a substitute.
Can I request a particular workplace or activity?
Yes, include it in your inquiry. Fiund will review sourcing feasibility, access, participant permissions, and capture constraints before discussing a suitable collection.
Is a dataset ready to download here?
This site is a sourcing inquiry point. Availability, samples, permitted uses, and delivery details are established for the specific project, rather than promised by the category page.