Now recruiting capture teams worldwide. Get paid to record first-person video of real work.

Apply to capture

Task row

Egocentric Chilled Display Case Restocking Data

Restock a chilled display case: one catalog row names this retail floor job, recorded by a multiview rig while the operator uses handheld grippers, against a spec you write, for egocentric display case restocking data, first-person chilled case restock video, a handheld gripper retail restocking dataset and fast, on-demand custom retail data collection.
Catalog row
"Restock display case" (Easy)
Environment
Retail Floor Set; Available on request
Capture
Multiview rig and handheld grippers
License
Perpetual, buyer-owned

Requirements

What does display case restocking data need?

01

One named job, written as its own task

The catalog names a single row, so a spec gives the job a start state, an end state and accept criteria rather than leaving it inside a generic retail label.

02

The capture mode the row actually uses

The row is recorded by a multiview rig while the operator uses handheld grippers, so the demonstrations map onto a robot end effector. The site does not state head-mounted capture for it.

03

Hands working in and around a case

The site pairs 3D hand pose, 21 joints per hand with a per-joint visibility flag, with contact events, so a hidden finger is flagged rather than silently guessed.

04

Depth that is honest about reflective surfaces

The site states that wet and reflective surfaces degrade active-stereo depth, and that depth confidence maps mask low-confidence regions. A spec should say how depth is to be used.

05

Rights you can ship on

A perpetual, buyer-owned commercial license backed by a documented consent chain, with faces, plates and screens blurred at ingest.

Definition

What is egocentric chilled display case restocking data?

Egocentric chilled display case restocking data is real-world video of a person restocking a chilled display case, the job the site task catalog words as "Restock a chilled display case." The footage shows the hands, the items being restocked and the case they go into. Alongside the video, each episode follows the Firsthand episode schema, with depth, 3D hand pose, contact events and frame-accurate action segments aligned on one hardware clock.

Teams search for it under several names, including egocentric display case restocking data, first-person chilled case restock video, a handheld gripper retail restocking dataset and custom retail data collection. They all describe demonstrations of the same job, collected to a written spec. This page covers only what the Firsthand site states about it, and marks everything else as Not stated.

Task

Which catalog row covers chilled display case restocking?

One row in the site task catalog covers this work. It is "Restock display case", described as "Restock a chilled display case.", in the Retail Floor Set environment under Retail & Consumer Goods. The catalog rates it Easy, lists its capture as Multiview rig + handheld grippers and lists its status as Available on request.

Anything outside that row is Not stated. The row does not say what goes into the case, how the case is laid out, how it is opened or closed or in what order the work happens, and it names no case type, product, brand or temperature, so none is claimed here. If your policy needs one of them, you write it into the spec and we scope it, because the catalog rows are a starting vocabulary, not a fixed menu.

Status

What does Available on request mean for this row?

The site defines Available on request as work that is "Scoped and recruited to spec when a buyer asks for it." The row is not listed as Active collection or Ready to deploy, so the site does not say that a display case capture stream is running today, or that contributors and site access are already set up for it. A first batch is therefore quoted against your spec rather than drawn from a stream that already exists, and nothing on this page implies delivered data.

The site lists an off-the-shelf retail shelf stocking egocentric dataset as Shipping, with a chilled case among its environments, but it does not tie that dataset to this catalog row. Where the site states what is in stock, it says off-the-shelf covers what is already in stock and custom collection covers the rest: custom collection at scale through a network of thousands of capture providers, for a consistent, recurring data flow to labs.

Environment

Where is it captured, and is it a real store or a staged set?

The catalog illustrates the Retail Floor Set with "Shelving, a checkout, a garment rail and a pack bench." The site describes the capture mode of this row as a staged working set, so the row is not described as a live store. That differs from the retail robotics training data page, which describes capture on live retail floors with head- and wrist-mounted cameras, and from the Retail Store -- Back of House rows, which list a head-mounted phone or a multiview rig.

The site does not describe the display case itself. Its size, layout, number of shelves and whether it has doors are Not stated, and the site names no store, retailer or building. It does not state the number of sites or the regions, and capture at your own facility is listed as not offered. Environment, access and schedule are agreed in the written spec.

Capture setup

How is the display case job captured?

The catalog states one capture mode for the row: a multiview rig with handheld grippers. The site describes it as a staged working set recorded by a multiview rig while the operator performs the task with handheld grippers, so demonstrations map onto a robot end effector. The site also states that every sensor in a multiview rig, meaning the head camera, both wrist cameras, depth and the IMU, is hardware-triggered from a single PTP grandmaster clock.

The site does not state a head-mounted capture mode for this row, so this page does not claim one, and it does not claim that a stereo pair is running on it. Stereo capture is a general Firsthand offer that is agreed in the skill spec. Camera count, resolution, frame rate, case layout and the items restocked are Not stated, so raise any of them in the spec. The word egocentric here follows the site page family; the capture mode is the one the row lists.

Compare

How is this different from the shelf restocking and grocery pages?

The site already has a stereo retail shelf restocking page that groups nine restocking, facing and stock-staging rows, this one among them, and treats them for stereo and MCAP delivery. It also has a retail robotics training data page on shelf-ops skills captured on live retail floors, a grocery put-away page about putting groceries away at home, and a grocery bagging page about scanning and bagging at a checkout. This page narrows to the single "Restock display case" row and describes the capture mode that row uses. The comparison table below sets each neighbor against this page.

A buyer who needs shelves, cartons, stocking carts and facing starts from the shelf restocking page. A buyer who needs a broader retail skill list captured on live floors starts from the retail robotics page. A buyer whose policy ends at a restocked chilled display case in a staged working set starts here.

Limits

What does this page deliberately not claim?

It does not claim that display case data is being collected today or has been delivered, because the row is Available on request and the site lists no volume, hours or episodes for it. It does not claim head-mounted capture or a running stereo pair, because the site states neither for this row. It does not claim that any named model was trained on Firsthand data, and it states no results from robots.

It also names no case, product, packaging, retailer or temperature, and it describes no steps as instructions, because the catalog row gives only its one-line description. Where a buyer needs any of those, the spec is the place to define them, and the quote that follows is written against that spec. Keeping the page to the stated row is what lets a reader check every sentence against the live catalog and the environment page.

Episode contents

What does each display case episode contain?

Each episode follows the Firsthand episode schema: ten streams synced under 2 ms, with head and wrist RGB, depth, 3D hand pose with per-joint visibility, contact and grasp events, and frame-accurate action segments. The site states hand pose as 21 joints per hand with a per-joint visibility flag, so a finger hidden behind an item is flagged as hidden rather than silently guessed, and contact events mark when a hand meets an item. The only language data the site states is one task instruction per episode.

Object labels use the terms your spec sets. Items, case types, packaging and layouts are Not stated. Task durations, cycle times and the number of episodes are Not stated too. Labels such as full, faced or restocked are Not stated, as are certifications, planogram compliance and robot results. The site makes no safety-compliance or food-handling claim for this row.

Why it is hard

Why is a chilled display case hard to capture well?

The catalog rates "Restock display case" Easy, so the site does not call it hard, and this page makes no claim about robot results. What the site does say is where related capture is difficult. It states that wet and reflective surfaces degrade active-stereo depth, and that depth confidence maps mask low-confidence regions. Whether a chilled display case or its contents behave that way is Not stated, so a spec should expect the same care.

In its off-the-shelf retail shelf stocking dataset entry the site also says that chilled-case capture is time-boxed to protect stock, so those episodes are shorter. Whether that applies to this row is Not stated. On the hands, the site notes that restocking puts fingers behind products and inside boxes, which is why close-range wrist views and hand pose with per-joint visibility are paired with it. The site also states that a robot or humanoid perceives from a head- or wrist-mounted camera, not a static overhead shot; this row is recorded by a multiview rig, so the geometry a buyer receives is the one the rig and the spec define.

Scoping

How do you split display case work into spec tasks?

A good spec treats the job as its own task. It names a start state, an end state and a set of accept criteria, and it says where one episode ends. The row words the job as one line, so which items, how many and in what order are yours to define, and the site states no episode length for it. Naming the end state keeps action segments consistent across contributors and lets you ask for more of this job without re-collecting another.

The same environment lists two neighboring restocking rows that a spec can chain with this one if your policy needs a longer run: "Shelf restock + face" and "Unbox + stage stock". Both are rated Easy, use a multiview rig with handheld grippers and are Available on request. Whether such chained sequences are collected as one episode or several is agreed in the spec.

Custom vs existing

When is a custom collection better than the off-the-shelf retail dataset?

The site describes the existing retail shelf stocking dataset as a fixed, already-captured corpus, available under its published terms. A custom collection is scoped to your written spec instead, with the exact job, capture mode and accept criteria your policy needs, captured on demand rather than chosen from a fixed catalog.

The retail robotics page puts the choice plainly: if the existing dataset already covers your task list, it ships faster, and if it does not, a custom collection fills the gap. The site does not state that the existing dataset covers the "Restock display case" row, or that it was captured with handheld grippers, so a policy that needs this row on that capture mode is a custom request.

Delivery

How is display case data labeled and delivered?

Delivery formats are RLDS, LeRobot and WebDataset, plus HDF5, zarr and Rerun .rrd. These episodes use the same schema as every other Firsthand collection, so a loader that reads another Firsthand dataset reads this one without changes. The format table below sets each format against what its official documentation states.

Every batch is validated against signed accept criteria before it ships, with a reject log, and delivery is rolling, per validated batch. Rejected items are recaptured at our cost, and billing is per validated hour.

Privacy

How are people in frame protected?

Every episode links to a signed contributor release through a documented consent chain, and the consent record is part of the delivery. A vision model runs on every frame at ingest, flagging faces, heads, licence plates and any legible screen or document, and those regions are irreversibly blurred. A human reviewer confirms the result before footage ships.

If the reviewer is unsure a region is fully covered, the episode is held or discarded, never downgraded and shipped. In its retail shelf stocking dataset entry the site adds that every bystander is blurred at ingest and that clips where a bystander cannot be de-identified are discarded. Who may be in frame for this row is agreed in the spec, and the site states no other privacy, compliance or legal terms for this work.

Fast · on demand

How fast can a custom retail dataset be collected?

Where coverage is already deep, the first validated batch lands in a median of 48 hours from signed spec. Thin environments typically take three to four weeks, because vetting operators there is the bottleneck, and we quote that up front. The site does not state how deep coverage is for this row, and it is Available on request, so plan on a quote rather than assuming the 48 hour figure.

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. Those assume capture coverage is already in place. On-demand top-ups run against the same spec and accept criteria after each training run. A batch size and a later-batch turnaround are Not stated, so both are quoted against your spec.

The spec

How do you start a pilot?

You describe the job, the items and the case as your policy meets them, and the sequence you need. We turn that into a written spec covering tasks, environments, capture tier, annotation depth and accept criteria, then quote reach, timeline and price before any capture begins. This is the custom egocentric data collection route that every Firsthand task family follows.

A Spec Pilot, a small and fast first batch, can prove the spec and accept criteria before you scale. Every collection ships under a perpetual, buyer-owned commercial license, with a reject log and signed accept criteria on every batch.

At a glance

What you get.

Every collection runs through the same seven-stage end-to-end custom collection pipeline. End-to-end custom data collection is a managed service that takes an AI data need from problem to owned dataset in a single accountable pipeline.

What the site states

What does the site state about chilled display case restocking?

Where this job or its neighbors appear on the site, what each place states and how this page uses it.
Where it appearsWhat the site statesUsed on this page as
Task catalog, Retail Floor SetA "Restock display case" row, "Restock a chilled display case.", rated Easy, recorded with a multiview rig and handheld grippers, Available on requestThe one stated task
Environment examples"Shelving, a checkout, a garment rail and a pack bench"The stated setting
Capture mode definitionsA multiview rig records a staged working set while the operator uses handheld grippers, so demonstrations map onto a robot end effectorHow the row is recorded
Stereo retail shelf restocking pageLists "Restock display case" among nine restocking, facing and stock-staging rows, three of them in Retail Floor SetWhere the row sits in the wider restocking family
Retail robotics training data pageShelf-ops skills on live retail floors with head- and wrist-mounted capture, including chilled-case rotationA neighbor with a different capture mode
Off-the-shelf retail shelf stocking datasetListed as Shipping, with a chilled case among its environments; chilled-case capture is time-boxedContext only; not tied to this row

Task rows

Which catalog rows cover display case restocking?

The row as the catalog words it, with the two neighboring restocking rows from the same environment shown for context.
RelevanceCatalog task and descriptionSkillCaptureStatus
This pageRestock display case: restock a chilled display caseEasyMultiview rig + handheld grippersAvailable on request
Neighbor, not covered hereShelf restock + face: restock a shelf and face productsEasyMultiview rig + handheld grippersAvailable on request
Neighbor, not covered hereUnbox + stage stock: open a carton and stage stockEasyMultiview rig + handheld grippersAvailable on request

Compare

How does this page differ from the neighboring pages?

Related pages on this site and what each one covers.
PageWhat it covers on this siteHow this page differs
Stereo retail shelf restocking dataNine restocking, facing and stock-staging rows across four retail environments, treated for stereo and MCAP deliveryThis page narrows to the one "Restock display case" row and its multiview rig with handheld grippers setup
Retail robotics training dataShelf stocking, facing, restocking and picking on live retail floors, with head- and wrist-mounted capture and a named skill listThis page covers one catalog row recorded as a staged working set, not a live store
Egocentric grocery put-away dataTaking groceries out of bags, carrying them across a kitchen and placing each item where it belongsA kitchen task, not a retail display case
Egocentric grocery bagging dataScanning items, taking payment and bagging goods at a checkoutA checkout task, not restocking a case
Retail shelf stocking dataset (off the shelf)An already-listed dataset, status Shipping, with a chilled case among its environmentsCustom collection to your spec; the dataset is not tied to this row or to handheld grippers

Streams and labels

What does each episode carry, and what is not stated?

Streams and labels the site states for Firsthand custom collections, and display case fields the site does not state.
ItemStatus on this siteWhy it matters for display case work
Head and wrist RGB, depthStated; ten streams synced under 2 msKeeps the case, the items and both hands on one timeline
3D hand pose, 21 joints per hand, per-joint visibilityStatedFlags a hidden finger when an item or the case covers a hand
Contact and grasp eventsStatedMarks when a hand meets or releases an item
Action segmentsStated; frame-accurateSeparates the steps of the job inside a session
Task instructionStated; one per episodeThe only language data the site states
Stereo pair on this rowNot stated for the row; stereo is a general offer agreed in the skill specRaise it in the spec if your policy needs it
Case type, size, layout, items, packaging, brands, temperaturesNot statedDefined in the spec
Task durations, cycle times, throughput, number of sites or episodesNot statedDefined in the spec
Success or quality labels, planogram compliance, certifications, robot resultsNot statedDefined in the spec if needed

Formats

Which formats carry display case data?

Delivery formats stated on the site, set against what each format official documentation states.
FormatWhat the official documentation statesUsed for here
RLDSAn ecosystem of tools to store, retrieve and manipulate episodic data; a dataset is retrieved as episodes, and each episode holds a dataset of steps (RLDS repository README)Episode-based training loaders
LeRobotLeRobotDataset v3.0 is a standardized format for robot learning data, including multi-camera video, with many episodes stored per Parquet or MP4 file (LeRobot documentation)LeRobot dataset format and tooling
WebDatasetTar files in which files that belong together share a basename, with shards numbered in sequence (WebDataset README)Sharded streaming for large training runs
HDF5 and zarrStated by Firsthand onlyArray storage for custom pipelines
Rerun .rrdStated by Firsthand onlyMulti-stream synchronized playback for review

Example footage

What does egocentric stereo footage look like?

One real clip already published on this site. It is a format and rig example: a head-mounted stereo capture with hands in frame. It does not show a display case or restocking, and it is not an example of the multiview rig with handheld grippers this row uses.

egocentric_stereo_sample.mp4 · head-mounted stereo · hands in frame · 1920 × 1080 web preview · rig example, not a display case clip

No chilled display case or retail restocking clip is published on the site, so this is the nearest first-person clip. Request a sample to see the delivered files. See stereo sample page, stereo + wrist-cam example videos, request an egocentric data sample.

FAQ

Egocentric display case restocking data, answered.

What is egocentric chilled display case restocking data?
Real-world video of a person restocking a chilled display case, captured against a written spec with depth, 3D hand pose and frame-accurate action segments.
Which task does the site state?
One Retail Floor Set catalog row: "Restock display case" ("Restock a chilled display case."), rated Easy, recorded with a multiview rig and handheld grippers, and listed as Available on request.
Is it captured head-mounted or in stereo?
The site does not state a head-mounted mode for this row. It describes it as a staged working set recorded by a multiview rig while the operator uses handheld grippers. Stereo is a general offer agreed in the skill spec.
Is this data available now?
The row is Available on request, which the site defines as scoped and recruited to spec when a buyer asks for it. It is not listed as Active collection or Ready to deploy, so a first batch is quoted against your spec.
Does it include case types, brands, temperatures or timing?
Not stated. The row names the job only, and everything else is defined in the spec.
How are people in frame handled?
Faces, heads, licence plates and legible screens are irreversibly blurred at ingest and a reviewer confirms the result. Who may be in frame is agreed in the spec.
How fast can a custom retail dataset be collected?
Where coverage is deep, the first validated batch lands in a median of 48 hours from signed spec, and thin environments typically take three to four weeks. The row is Available on request, so timing is quoted against your spec.

Sources

Where are the delivery formats documented?

Request a custom dataset.

Tell us the chilled display case job your policy needs to handle, the items and the case as it meets them, and the sequence. We will quote reach, timeline, and price against a written spec, scoped and recruited to that spec.