Egocentric Promo Display Assembly Data

Egocentric promo display assembly data is first-person video of one retail task: building a small promotional display. It is recorded from the wearer viewpoint, close to what a robot onboard camera would see while setting up a display, with depth, 3D hand pose, contact events and frame-accurate action segments aligned on one hardware clock. Teams searching for retail display robot training data, egocentric promotional display assembly video or a visual merchandising setup dataset are usually looking for this.

On this site the data is human capture: a paid, consented contributor performs the task from a written skill spec. It is not robot or teleoperation data. It is custom retail data collection rather than a ready-made library, with the task written down, collected on demand and delivered as validated episodes.

The task catalog states one row for this task: Assemble promo display, in the Retail Floor Set environment. The environment is described in the catalog as "Shelving, a checkout, a garment rail and a pack bench." The row is rated Easy, its capture mode is Multiview rig + handheld grippers, and its status is Available on request.

This page claims no delivered hours, and no promo display footage is published on the site. The first table below gives the row exactly as the catalog words it.

The row reads "Build a small promotional display." That is the whole description. The site names no display type, no fixture, no product, no signage, no retailer, no store and no assembly time, and it does not say what a finished display looks like or how many items it holds.

Because the site states none of those details, this page does not either. Which parts are in scope, where an episode starts and ends, and what counts as finished are agreed in the written spec. The second table below sets each item against what the site does and does not state.

The row uses a multiview rig with handheld grippers. The operator works with the grippers in hand, so the demonstration maps onto a robot end effector, and the multiview rig records the task from several synchronized viewpoints. The site states no camera count, resolution or frame rate for this row, so this page makes none.

Each episode follows the episode schema used across Firsthand custom collections: ten streams synced under 2 ms, including head and wrist RGB, depth, 3D hand pose with per-joint visibility, contact and grasp events, and frame-accurate action segments. The site does not say how the ten-stream schema maps onto the multiview rig with grippers for this row, so confirm the channels you need on the Spec Pilot. The third table below lists each stream and label.

Each attempt is delivered as one episode. The site does not set an episode length or count, so the skill spec does. Action segments divide the recording into spans, each carrying start_ns, end_ns, verb and noun, so a build can become labeled steps as the spec defines them rather than one long clip. Contact events mark when a hand or gripper meets a part of the display.

Two machine-checked rules apply to every custom collection. An episode is rejected if the manipulated object is out of frame for more than 15 percent of it, and if more than 35 percent of depth pixels in the manipulation region are invalid. A display has many parts, so the spec should say which object counts as manipulated at each step.

Several retail pages cover other work in the same environment. The fitting room and apparel merchandising page covers apparel rows, the price tagging and shelf signage page covers tags and signage, the unboxing and unpacking page covers getting into cartons, and the stereo retail shelf restocking page covers stereo-rig capture of shelf restocking. None of them covers building a promotional display.

The retail robotics training data page is the broader hub for shelf stocking, restocking, facing and picking. The fourth table below sets the pages side by side.

No. No dataset on the off-the-shelf pages is listed for promo display assembly. The closest retail dataset on the site covers facing, restocking, price changes and planogram resets, which is a different task, so this is custom collection. The row is Available on request, so the work is scoped and recruited to your written spec.

Every episode links to a signed contributor release through a documented consent chain, and the consent record is part of the delivery. Faces, plates and screens are detected and irreversibly blurred at ingest, and a reviewer confirms the result before anything ships. Clips that cannot be de-identified are discarded. The anonymization record is delivered with the data.

The site names no store, chain or retailer for this row and does not say that readable signs or printed text on a display are blurred. If the display carries text that matters to you, raise it before the spec is signed. Capture at your own facility is not offered, so this page makes no claim about where episodes are recorded.

Delivery formats are RLDS, LeRobot and WebDataset, plus HDF5, zarr and Rerun .rrd. The first three are the training-loader formats; HDF5 and zarr carry arrays for custom pipelines, and Rerun .rrd gives a multi-stream synchronized playback file for review. The fifth table below sets each against what its official documentation states. The site states one task instruction per episode as the language data for a Firsthand capture.

Every batch is validated against signed accept criteria before it ships, with a reject log, and delivery is rolling, per validated batch.

The site states a median of 48 hours from signed spec to first validated batch where coverage is deep, and three to four weeks where it is thin and contributors need to be recruited first. This row is Available on request, and the site does not say whether its coverage counts as deep or thin, so ask which timeline applies when you request the quote.

The site also states program milestones: a pilot reaches its first delivery in two to three weeks, and a standing program reaches full rate in four to six weeks. On-demand top-ups run against the same spec and accept criteria after each training run.

You describe the work your policy needs: which display, its parts, the start state and the end state, and the setting. We turn that into a written spec covering the steps, the environment, the capture setup, annotation depth, accept criteria and delivery format, then quote reach, timeline and price against that document before any capture begins. The last table below lists each element.

A Spec Pilot, a small and fast first batch, can prove the spec and the accept criteria before you scale volume. Billing is per validated hour, a reject log ships with every batch, and rejected items are recaptured at our cost.