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

Apply to capture

Task family

Egocentric Postal Sorting Data

Sorting letters and parcels by route into pigeonholes and trays at the sorting bench, captured first-person against a spec you write, for mail sorting robot training data and fast, on-demand custom logistics collection.
Streams
Ten, synced under 2 ms
Task status
Available on request
Billing
Per validated hour
License
Buyer-owned

Requirements

What does postal sorting data need?

01

A named sorting task

Sorting letters and parcels by route into pigeonholes and trays is one stated sequence, so a spec names it and says where each item type goes.

02

Both hands, the item and the destination in frame

A head-mounted view keeps both hands, the letter or parcel and the pigeonhole or tray in view through each placement.

03

Hand pose through each placement

3D hand pose with per-joint visibility carries how a hand picks up a letter or parcel and places it, and shows where a hand is hidden.

04

Step boundaries on the same clock

Frame-accurate action segments mark where one placement ends and the next begins, aligned to video and depth.

05

Rights you can ship on

A perpetual, buyer-owned commercial license backed by a documented consent chain.

Why it matters

What is egocentric postal sorting data, and why does it need its own page?

Egocentric postal sorting data is real-world, first-person video of mail sorting: taking letters and parcels and placing them by route into pigeonholes and trays at a sorting bench. It is recorded from the sorter viewpoint, with depth, 3D hand pose, contact events and frame-accurate action segments aligned on one hardware clock.

The task is a bench task. Nothing is carried to a door and nothing is handed to a recipient, which separates it from delivery. The package delivery page covers the delivery end of the chain, and this page covers the sorting that the catalog lists beside it.

The catalog row

Which catalog row covers postal sorting, and what is its status?

The task catalog has one row for it. In the Last-Mile Delivery & Courier environment, under the Logistics & Warehousing sector, the Postal sorting row reads "Sort letters and parcels by route into pigeonholes and trays at the sorting bench." It is rated Easy and recorded with a head-mounted phone or a multiview rig.

The row is listed as Available on request. That means it is quoted against your spec, and this page does not claim active or completed collection for it. The site states no capture timeline for the row beyond a quote.

What the row names

What does the row name, and what does it leave out?

The row names letters, parcels, a route, pigeonholes, trays and a sorting bench. Those are the only items and steps this page treats as stated.

Letter and parcel sizes, the number of routes, pigeonholes or trays, the bench layout, whether items are read for a route code, and who the sorter is are Not stated. They are set in the written spec. The page names no postal carrier and makes no sort-rate, accuracy or throughput claim.

Neighbouring pages

How is this different from package delivery, parcel handling, tote induction and mailroom data?

Four existing pages sit nearby, and none of them has postal sorting as its main topic. The package delivery page centres on carrying a parcel to the door and handing it to a recipient, and names Postal sorting only as a neighbouring row. The box and parcel handling page covers palletizing, carrying, sealing and scanning, and lists Postal sorting among the rows that sort by route.

The tote induction page covers putting totes and items onto a conveyor, and its page text names no postal row. The mailroom page centres on the Mail handling row in Office & Professional Services and on document work, with the Postal sorting row as one of its rows. The table below sets them side by side.

Environments

Where is postal sorting captured?

The catalog places the row in the Last-Mile Delivery & Courier environment. The site names no postal facility, sorting office, city, carrier or courier, and it does not state capture at a customer own site for this row.

Environment, access and schedule are agreed in the spec. The number of sorters or sites is Not stated.

Capture setup

How is sorting captured from a first-person view?

A head-mounted phone records the first-person view of real work on site, and a multiview rig adds synchronized third-person angles when the spec asks for them. Both are the capture modes the catalog lists for the row.

The site does not state a camera count, resolution or frame rate for postal sorting, so these are raised in the spec. Which of the two modes a postal sorting order uses is also Not stated.

Streams and labels

What does each postal sorting 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. For sorting, hand pose tracks how a hand takes a letter or parcel and places it, contact events mark when fingers meet an item, and action segments mark each placement.

Object labels use the terms your spec sets. A route label per item, a sorted-correctly or task-completion label, and force or tactile data are Not stated. They are defined in the spec if your policy needs them.

Quality

How is the quality of sorting episodes checked?

Every batch is validated against signed accept criteria before it ships, with a reject log, and rejected items are recaptured at our cost. Billing is per validated hour, not per raw hour recorded.

Per-joint visibility in the hand pose shows frames where a hand is hidden, for example behind a tray or another item, so a reviewer and a loader can see where the pose is less certain. The site states no sort-accuracy figure, so accuracy is not claimed here.

Delivery

How is postal sorting data labeled and delivered?

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

Delivery is rolling, per validated batch.

Privacy

How is privacy handled?

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. The anonymization record is delivered with the data.

The site states no rule specific to mail contents, addresses or labels on letters and parcels, so this page makes no claim about them. If your spec involves such items, raise it before capture.

Fast · on demand

How fast can a custom postal sorting 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 Postal sorting row is Available on request, so the site does not state which of the two timelines applies to it. It is quoted against your spec, and on-demand top-ups run against the same spec and accept criteria after each training run.

The spec

How do you start a pilot?

You describe the letters, parcels, routes and sorting layout your policy needs. 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.

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

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.

The catalog row

What does the catalog state about postal sorting?

The Postal sorting row as listed in the task catalog, and what the site does and does not state about it.
FieldWhat the site statesUsed on this page as
TaskPostal sortingThe page topic
DescriptionSort letters and parcels by route into pigeonholes and trays at the sorting benchThe only stated items and steps
Environment and sectorLast-Mile Delivery & Courier, Logistics & WarehousingWhere the row is listed
Skill ratingEasyQuoted as listed
Capture modeHead-mounted phone or multiview rigThe stated capture modes
StatusAvailable on requestStated plainly; no active or completed collection is claimed

Comparison

How does this page differ from the neighbouring pages?

Four existing pages that touch postal or parcel work, their main topic, and how each treats the Postal sorting row.
PageMain topicPostal sorting row
Egocentric postal sorting data (this page)Sorting letters and parcels by route into pigeonholes and trays at the sorting benchThe main topic
Egocentric package delivery dataSorting parcels in a van, carrying them to the door and handing them to a recipientA neighbouring row, kept out of delivery
Egocentric box and parcel handling dataPalletizing, carrying, sealing, scanning and sorting boxes and parcelsOne of several rows that sort by route
Egocentric tote induction and conveyor loading dataInducting totes and loading items onto a conveyorNot named on that page
Egocentric mailroom and document handling dataMail handling in Office & Professional Services, plus document workOne row among the mail and document rows

Streams and labels

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

Streams and labels the site states for Firsthand custom collections, and postal sorting fields the site does not state.
ItemStatus on this siteWhy it matters for postal sorting
Head and wrist RGB, depthStated; ten streams synced under 2 msKeeps the item, the bench and the hands on one timeline
3D hand pose, per-joint visibilityStatedCarries how a hand takes and places an item, and where a hand is hidden
Contact and grasp eventsStatedMarks when fingers meet a letter or parcel
Action segmentsStated; frame-accurateSeparates one placement from the next
Letter and parcel sizes, bench layoutNot stated; the row names letters, parcels, pigeonholes, trays and a bench onlyDefined in the spec
Number of routes, pigeonholes or traysNot statedDefined in the spec
Route label per item, sorted-correctly or completion labelNot statedDefined in the spec if needed
Sort rate, accuracy or throughputNot statedNo such claim is made on this page
Force or tactile dataNot statedDefined in the spec if needed
Postal facilities, carriers or regulationsNot statedNo such claim is made on this page

Formats

Which formats carry postal sorting data?

Delivery formats stated on the site, set against what each format official documentation states where the documentation was read.
FormatWhat the official documentation statesUsed for here
RLDSA 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, split into numbered shards (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 footage of hands-on work and stereo capture look like?

Two real clips already published on this site. The first is a household chores clip of an operator carrying a cordless vacuum down a staircase into a living room, a first-person hands-on clip hosted on the site. The second is a format and rig example: a head-mounted stereo capture with hands in frame. Neither shows postal sorting.

sample-egocentric-capture.mp4 · household chores · carrying a vacuum between floors · fisheye, 960 × 540 web preview · not a postal sorting clip
egocentric_stereo_sample.mp4 · head-mounted stereo · hands in frame · 1920 × 1080 web preview · rig example

No postal sorting clip is published on the site, so these are labelled as format and rig examples. Request a sample to see delivered files. See stereo sample page, stereo + wrist-cam example videos, request an egocentric data sample.

FAQ

Mail sorting robot training data, answered.

What is egocentric postal sorting data?
Real-world, first-person video of a person sorting letters and parcels by route into pigeonholes and trays at a sorting bench, captured against a written spec with 3D hand pose, depth and action segments.
Is postal sorting collection active or completed?
No. The Postal sorting row in the Last-Mile Delivery & Courier environment is rated Easy and listed as Available on request, so it is quoted against your spec. This page claims no active or completed collection for it.
Does the site state sort rates, accuracy, postal regulations or named carriers?
No. The row names letters, parcels, a route, pigeonholes, trays and a sorting bench only. Sort rates, accuracy, throughput, postal regulations and carriers are Not stated, and this page makes no claim on them.
How is this different from package delivery and mailroom data?
Package delivery centres on carrying a parcel to the door and handing it to a recipient. Mailroom data centres on the Mail handling row and document work. This page centres on the Postal sorting row at the sorting bench.
Which formats can the data be delivered in?
RLDS, LeRobot and WebDataset, plus HDF5, zarr and Rerun .rrd. Every episode uses the same Firsthand episode schema, with ten streams synced under 2 ms.
How fast can a custom postal sorting 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 its timeline is quoted against your spec.

Sources

Where are the delivery formats documented?

Request a custom dataset.

Tell us the letters, parcels, routes and sorting layout your policy needs to handle. We will quote reach, timeline, and price against a written spec, with a first validated batch in a median of 48 hours where coverage is deep.