Dataset program

Build retail robots that understand products, shelves, fabrics, packaging, and real customer-facing workflows.

Real retail human demonstration data for garment folding, stocking, product handling, packaging, cake decoration, and point-of-sale workflows.

Video context. A real retail workflow showing product handling and fine-motor task execution for robotics dataset evaluation.

Built for model teams

Useful coverage, documented clearly.

Every program is scoped around the task decisions your team needs to make, then delivered with the context needed for review, training, and audit.

Garment folding and fabric handling

Shelf stocking, sorting, and product presentation

Packaging and point-of-sale workflows

Fine-motor food and retail preparation tasks

Capture specification

EnvironmentsRetail shops, counters, displays, stock areas, and bakeries

TasksFolding, stocking, sorting, packaging, decorating, and showing

ViewsEgocentric clips with optional exocentric workspace context

DeliveryVideo clips, action and object labels, timestamps, and QA notes

Common questions

Are retail workers identifiable in the data?

Privacy review and anonymization are part of the delivery workflow, with consent collected before release.

Can retail data cover multiple store formats?

Yes. Programs can define store layout, product category, task variation, and capture volume.

Share the task, objects, environment, target hours, and schema. We will recommend the smallest useful sample.