Why this is hard
A model can read every sentence ever published and watch every video ever uploaded, and still not know how much force a drawer needs. Physical intelligence is not written down anywhere; it is performed, by people, and then it is gone.
Language models
Already collected, already public. The corpus existed before the models did.
Vision models
Images and video scraped from an internet that was already there. Again, found rather than made.
Physical AI
Nothing equivalent exists. Egocentric capture of ordinary human work has to be recorded on purpose, by someone, with consent.
The first two corpora were found. The third has to be made, and that asymmetry is the point.
Check it yourself
We don't sell stock footage. We capture to your brief, and every recording we deliver carries a consent receipt you can verify yourself. Check a receipt
Every capture session writes one row into an append-only ledger. Eleven fields join with a pipe, that string is hashed, and the hash of the row before it is one of the eleven; so changing any earlier row breaks every hash after it. The record below is an example, because a real one names a real person. The digest is not an example: it is produced by the same function that runs in production.
The eleven fields, in the order they are joined
The pre-image: those eleven, joined by |
loading…SHA-256 of that string
loading…That is the entire claim, and it does not require us to be honest, only to be consistent, which is checkable. So check it: the button below hashes the pre-image above in your browser, with the Web Crypto API your browser ships, and compares the result to the digest we published.
One recording
Five layers, one real recording, and every figure in them computed from the file itself. Each step links to the page that sets it out in full.
Faces the detector finds — blurred automatically, server-side at ingestion, before any human review.
—01 · Capture
measured
For a commissioned brief, the task, the environment and the rig are fixed before anything is recorded. Drag the handle to move between the raw frame and the annotated one.
Consent receipt: none — recorded before the consent flow
02 · Redact
Face redaction runs at ingestion, server-side, before any human review. If the face detector cannot load, the clip fails the stage rather than passing through.
03 · Segment
derived
The recording is divided into sixteen chunks and each is given a verb-noun code. On this clip that division is a fixed window of 120 frames, not a boundary found in the footage, and the codes come from a fixed pool rather than a model. The bars below sit at their true frame positions.
04 · Annotate
derived
Hand landmarks are computed every tenth frame — a three-hertz track, not thirty. Drag the timeline to move between sampled frames. The scrubber never interpolates: every position it shows is a row of the artifact, and there is nothing between two sampled frames to show.
05 · Deliver
measured
The finished record is content-addressed: rehash the annotation file you download and you get the digest below. The trajectory below is drawn complete rather than animated, because a path easing into place invents positions between the sampled ones.
By the numbers
Frames annotated end to end and published in full
1,843
measured
One real recording at 30 fps, with its annotation and its checksum.
Delivery containers
5
measured
Each has a writer and its own verifier, and every export is read back off disk before it ships.
Mean per-hand confidence
0.87
derived
The hand tracker's own confidence score, averaged over the kinematic track. The lowest is 0.51.
Clips specified across the catalogue
182,700
target
Across 6 capture domains. Each brief states its own rig, its annotation depth and where it stands.
Hours of egocentric capture, specified
1,505
target
Volume is agreed in a brief and recorded against it. Simulated scenes are catalogued separately.
Catalogue
Each domain below is a capture specification you can commission, and each ships against FH-Ego v1, so a dataloader you write once works across every delivery.
Tier 1 · Session
Scene, task label, frame count, frame rate, usability verdict, and the provenance block: clip id, content hash, cohort, QA verdict, and which fields were derived rather than measured.
Tier 2 · Action
Frame bounds per action segment, hand laterality (left / right / both), scene semantics, camera motion, a verb-noun action code and a dense natural-language description.
Tier 3 · Kinematics
21 3D joint landmarks per hand, camera-relative wrist position, frame-to-frame velocity, and a per-hand confidence; so you can filter on quality rather than trust it.
The catalogue also scopes agriculture and outdoor tasks, and pairwise preference ranking on policy rollouts.
Why egocentric
Teleoperation gives you a robot's own proprioception, and it costs one person driving one rig for every hour of demonstration it produces. Egocentric capture gives you human hands solving the task in the environment the task actually happens in, and it scales with people rather than with rigs. There is a real case against it (no joint torques, no action labels for free, a domain gap you have to close), and the published work arguing both sides is a category argument, not a Fourthuman result. We set it out on the method page.
Price
Three things move the number, and each is fixed in the brief.
What moves the number · 1
An Indian household kitchen and an active distribution centre are not the same shoot. Access, scheduling and how many people have to consent all change with the room.
What moves the number · 2
How long a session runs, how many of them, and how many Capture Partners the brief needs to reach the volume you are asking for.
What moves the number · 3
Tier 1 is session metadata. Tier 3 is 21 3D joints per hand with per-hand confidence, and it costs more to produce and more to verify. You pick the tier in the brief.
The structural reason a bespoke brief is affordable here at all is the India–US wage differential.
Looking for paid recording work? This form is for labs buying data. Register as a Capture Partner →
For robotics and AI labs
Describe the policy and the embodiment you are training it for. We scope the capture specification, price it against your volume and format, and come back with what we can commit to, including the parts we cannot. Our team reads every one of these.
What a brief costs is scoped before it is priced: how we price, and what moves the number.
FAQ
lerobot package), HDF5 (EgoDex-style paired .hdf5+.mp4, ALOHA / RoboMimic-lineage labs), and the open FH-Ego v1 JSON schema. Zarr v2 (UMI / diffusion-policy layout) and TFRecord (RLDS-style step schema; not tfds.load-compatible) are produced on request. Every export is read back by an independent verifier before delivery. We do not offer MCAP or WebDataset: MCAP is a robotics logging container no training dataloader consumes, and WebDataset is an internal sharding optimization, not a distribution format.