Egocentric Returns Processing Data

Egocentric returns processing data is first-person video of an associate handling returned items, recorded from their own point of view with synchronized sensor streams alongside. The camera sits on the head, close to what a robot’s onboard camera would see, so the frames show what the hands do with the parcel, the item and the tag. The same idea sits behind any returns processing robot training data or reverse logistics video dataset request, applied here as a custom collection scoped to your tasks.

The data is human capture: a paid, consented contributor wears the rig and performs a task from a written skill spec. It is not robot or teleoperation data. It is also not fixed-camera warehouse or store footage, because the viewpoint is the person doing the work.

The task catalog lists three rows about returns directly. "Returns Processing" sits in the Warehouse & Distribution Center environment, in the Logistics & Warehousing sector, which the catalog illustrates with an e-commerce fulfillment hub, a distribution center, a cold storage facility and a freight warehouse. It is described as opening returned parcels, inspecting and scanning items and routing them to restock or disposal. It is rated Easy and Ready to deploy.

"Process a return" sits in the Retail Floor Set environment, in the Retail & Consumer Goods sector, illustrated with shelving, a checkout, a garment rail and a pack bench. It is described as inspecting, re-tagging and re-shelving a return. It is rated Easy and Available on request.

"Checkout and returns" sits in the Retail Store -- Front of House environment, illustrated with a clothing store, an electronics shop, a furniture showroom, sporting goods, a bookstore and a department store. It is described as scanning purchases, processing payments and returns at the register and bagging or refunding items. It is rated Easy and Available on request.

No. "Returns Processing" and "Checkout and returns" use a head-mounted phone with a multiview rig. "Process a return" uses a multiview rig with handheld grippers instead. The first-person view of real work comes from the head-mounted phone, and the multiview rig adds synchronized third-person angles when the spec asks for them.

The site states no parcel types, item categories, return volumes, store formats, store chains or facility names for these rows.

Several rows touch returns without being returns processing. Damaged-goods sorting, where damaged goods are inspected and sorted with a handheld scanner and returns bins, is a sorting task. Restocking rows fill shelves from cartons or a cart. Checkout rows such as "Checkout and bagging" scan items and take payment at a grocery checkout. This page does not count any of them as returns processing.

Our sorting task page covers damaged-goods sorting and says that a return processing row on a retail floor set is a different task. Our box and parcel handling page covers moving and opening parcels in general. Both are linked below.

No. No catalog row names refund policy decisions, return authorization, fraud or abuse detection, grading an item’s condition, recording a return reason, working in a warehouse management or point-of-sale system, or contacting a customer. The coverage table below sets each of these against what the catalog actually lists.

If a program needs a task the catalog does not name, it is written into the skill spec, and whether it can be captured is confirmed when the spec is quoted rather than assumed.

Each attempt is delivered as one episode. On this site the standard episode carries action segments, 3D hand pose for both hands with a per-joint visibility flag, contact events that mark when a hand meets an item, an objects list, one task instruction, and per-frame instance masks stored at the depth resolution of 848×480.

For returns work, an object in the list might be a parcel, an item, a tag or a bin, as written in the skill spec. Whether an item was accepted or refused is not a stated stream.

Action segments divide an episode into start- and end-timestamped spans, each carrying start_ns, end_ns, verb and noun. For returns work, a segment might be one parcel opened, one item scanned or one tag replaced, as written in the skill spec. The site lists no label for the outcome of a step, such as whether an item was sent to restock or to disposal.

Validation runs against signed accept criteria written into the spec before capture, and a reject log ships with each batch. A human reviewer checks every episode. Only footage that clears the criteria is billed as a validated hour.

Delivery is in the format your stack already reads: RLDS, LeRobot, WebDataset, HDF5, zarr or Rerun .rrd, or raw streams with metadata, with readable converter source alongside. The license is perpetual, irrevocable and worldwide, buyer-owned and non-exclusive by default, with exclusivity available as a separate term. The executed agreement controls.

Returns look simple in the catalog: open, inspect, scan, route. The difficulty sits in the sequence. The hands go inside a parcel, so the contents and the fingers can be hidden. The same hands then hold a scanner or a tag, and the next step depends on what the inspection showed. That is why hand pose carries a per-joint visibility flag, why contact events mark when the hand meets the item, and why masks are recorded per frame. The challenge table below maps each difficulty to the stream or label this site states.

The destination of each item, restock, shelf, refund or disposal, is a decision. The site states no label for that decision or for item condition.

The catalog describes the head-mounted phone as recording the first-person view of real work. The site also states that Firsthand does not capture at a customer’s own facility.

Every contributor signs a written release, and each delivered episode links to it through consent artefact IDs. At ingest, automated detection finds faces, plates and screens and blurs them irreversibly, and speech is scrubbed of names, addresses and card numbers. A human reviewer confirms that no un-consented person is identifiable in any frame, and the anonymization record ships with the data. That matters at a register, where a customer or a screen can appear in frame.

Refund policy, fraud handling and certifications are not stated for these environments, and this page makes no such claim.

From signed spec to first validated batch, the published timeline is a median of 48 hours where coverage is deep and three to four weeks where it is thin and contributors need to be recruited first. The catalog marks "Returns Processing" Ready to deploy, meaning contributors and site access are already set up. "Process a return" and "Checkout and returns" are Available on request, so they start from the spec.

Describe the returns task and its order, the items or parcels involved, where each one ends up, what counts as a finished step, and the failure or variation cases your pipeline is missing. If you need something beyond the three rows, such as grading item condition or working in a return system, say so up front. That becomes a written skill spec, quoted for reach, timeline and price before any capture begins.

Most programs begin with a Spec Pilot, a small, fast first batch that proves the spec and the accept criteria on real footage before volume scales. Collections are billed per validated hour.