Egocentric multimodal dataset

Eye Tracking, RGB-D & Smart Glasses

Gaze, depth and smart-glasses capture for research that needs more than RGB.

In stock · Indicative volume195 hoursNon-exclusive licence
Eye Tracking, RGB-D & Smart GlassesIllustration · sample video pending

This data is in stock, but its de-identified public sample isn't published yet. Request a quote and we share the scenario list, measured capture specs and sample clips directly.

Representative scenarios

Typical tasks in this category. The exact scenario list, hours per scenario and capture specs are confirmed in your quote.

Price on requestSample pending

Eye-Tracked Meal Preparation

Slicing, stovetop cooking and plating with gaze fixations recorded.

slicestirplatelook
Price on requestSample pending

RGB-D Hand–Object Pose

Tool use and rigid-object handling recorded with depth for 3D hand–object pose.

grasprotateplaceuse tool
Price on requestSample pending

Smart Glasses at a Desk and at Home

Desk and multi-monitor work, hardware setup, and moving through a home with smart devices.

typeplug inwalkoperate
Price on requestSample pending

Toy Assembly and Disassembly

Repeated, standardised assembly and disassembly of multi-part toys at 60 fps.

pickfitscrewtake apart

Overview

Some models need to know where a person is looking or exactly where their hands are in 3D. These smaller sets add those signals to first-person video.

They include 28 hours of eye-tracked meal preparation with gaze fixations, 5.5 hours of RGB-D recordings of 3D hand–object pose and tool use in a robotics lab, 42 hours of smart-glasses recordings at 1408×1408 in VRS (desk and computer work, and moving through a home), and 120 hours of standardised toy assembly and disassembly at 60 fps in a controlled lab.

Recorded in the United States, Germany and Switzerland. Streams, formats and exact hours are confirmed in your quote.

Environments

  • Home kitchens
  • Robotics lab
  • Desks and homes
  • Controlled assembly lab

Built for

  • Gaze prediction and attention modelling
  • 3D hand–object pose estimation
  • Procedural and step-by-step task learning
  • Smart-glasses assistants

Licensing & availability

Inventory195 hours · indicative — exact hours confirmed in writing in your quote
Non-exclusive licenceYes — default for all in-stock hours
Exclusive licenceOn request, subject to existing licences
Commercial model trainingIncluded
Custom top-up collectionAvailable when you need more than stock or a new scenario
AnnotationRaw streams. Annotation to your spec is quoted.
ConsentWritten consent from every person filmed; documentation available during licence review
PrivacySamples de-identified before publication; delivery screening per licence terms
DeliveryOriginal video, gaze, depth or VRS files plus per-session metadata

FAQ

Why so few hours?

These are specialised research sets. We can record more with the same sensors to your task list.

Can I see samples?

Yes. Request a quote and we share samples of the streams you need.

More datasets

View all