Egocentric Cabinet Hardware Installation Data

It is first-person video of a person fitting hardware onto a cabinet or a piece of furniture, recorded from a head-mounted camera, with two wrist cameras for a closer view of the hands. The contributor is a paid, consented person following a written skill spec, not an actor and not a robot.

The point of the data is what the hands do across the whole job: marking, positioning a part, drilling, and driving the fixings. It is labeled so each step can be sliced out and trained on separately or as one longer episode. Where the site states nothing, a table says "Not stated" and the answer lives in the spec or the quote.

The task catalog lists two rows for it, both in the Carpentry & Furniture Workshop environment, which it describes as a joinery and furniture workshop with saws, sanders and finishing stations. One is "Drill and install cabinet hardware using a drill and hinge jig": "Position hinges with a jig, drill pilot holes and screw hinges and handles onto cabinets." The other is "Hardware Installation": "Mark, drill and fit hinges, handles and slides onto cabinets and furniture."

Both are Medium, both list a head-mounted phone or multiview rig, and they differ in status: the first is Active collection and the second is Ready to deploy. That is everything the catalog states, so the first table sets the two rows side by side and the rest of this page says "Not stated" wherever a buyer might expect a detail such as a cabinet type or a hardware model.

The wording is close but not the same. The hinge jig row names two tools, a drill and a hinge jig, and ends at screwing hinges and handles on. The Hardware Installation row names no tool, begins with marking and adds slides to the parts it lists, so it reaches furniture as well as cabinets.

A buyer can ask for either, or for both in one spec. The site does not say how the two overlap in practice or how much of one row is covered by the other, so the spec states which stages are wanted and how a hinge, a handle and a slide are each to be separated.

The two rows give a natural cut: mark the position; position the hinge with a jig; drill pilot holes; screw the hinges and handles on; fit the slides. A spec that says "install the hardware" gives you one long action segment. A spec that names those stages, and splits hinges, handles and slides, gives you a sequence you can slice by step and by part.

The step table sets each stage beside the episode field that carries it. The stages are the catalog’s own wording; how finely a spec splits them, and where each one starts and ends, is a spec decision and not a stated default. The catalog does not say how many hinges, handles or slides make up one episode.

The catalog ties both rows to the Carpentry & Furniture Workshop environment and lists a head-mounted phone or multiview rig. The wider site describes a head camera plus two wrist cameras, a depth sensor and an IMU, shown in the capture table below.

The site does not state which workshops host a given row, how access is arranged, whether the work is at a bench or on an installed cabinet, or how many contributors are available for it. The spec settles location and access, and the quote settles reach.

The catalog marks both rows Medium, the same level as sanding and finishing wood surfaces in the same environment, and one level below cutting and shaping wood panels, which is Hard. It does not say why a row sits at a level, so this page does not explain it beyond what the rows list: a marked position, a jig, pilot holes and fixings.

Active collection and Ready to deploy are the catalog’s statuses for the two rows. The catalog does not define them further, so ask what each means for your region and your timeline when you request the quote.

An episode is one contributor performing one task from the skill spec, recorded with ten hardware-synchronized streams that line up within 2 ms. The site documents a canonical episode schema, and every delivery format is a view of it.

Hand pose is stored as 21 joints per hand in metres in the head frame, with a visibility flag for each joint. Contact events record when a hand meets an object. The site does not say whether a hinge, a handle, a slide, the jig or the drill counts as the object, so settle it in the spec.

Action segments carry a start, an end, a verb and a noun from a taxonomy the site states as 97 verbs and 512 nouns, human-reviewed. Every episode links to its contributor’s signed release by a consent artefact ID, and each batch ships with a data card.

The unit of labeling is the action segment: one start, one end, one verb, one noun. A mark, a positioned hinge, a drilled hole and a driven screw can each be one segment if the spec asks for it. Instance masks track object IDs per frame at 848×480, which helps keep a hinge, a handle and the cabinet apart.

Every episode is machine-checked, then reviewed by a human against its skill spec, and rejected episodes are recaptured at our cost. The site states no label for hardware type, hole position, alignment, fit or task success, and no force, torque or tactile reading, so ask for any such label at the start.

The site describes a small set of failure modes and what it does about each. The first is specific to this work: the drill, the fingers or a jig can hide the spot being worked. Hands hidden from the camera are flagged by a per-joint visibility bit, not guessed, contact events mark when a hand meets an object, and per-frame instance masks keep objects apart.

Depth near close work ships with a per-pixel confidence map, so a buyer can see where it should not be trusted. The rig is re-solved against a checkerboard at the start and end of every session and accepted only under the 0.28 px RMS reprojection gate. Weak episodes are rejected and recaptured at our cost. The site states no further failure modes for this task, such as workshop lighting or dust, so those are raised in the spec.

Only what the site already states applies. Every contributor signs a written release, and each episode links to it. Faces and screens are detected and blurred irreversibly, a human reviewer confirms that no un-consented person is identifiable, and footage with bystanders who cannot be de-identified is discarded rather than shipped.

The site does not state how other people in a shared workshop are handled beyond that general rule, so settle it in the spec before capture. This is not legal advice, and the executed agreement controls.

The egocentric woodworking and carpentry page covers the whole workshop environment, with eleven rows including both hardware rows. The egocentric furniture assembly page covers assembling and joining furniture and lists Hardware Installation as a neighbouring row. The egocentric drilling and mounting page covers ten rows across five environments, including both hardware rows.

This page takes just the two hardware rows and goes deeper on their steps, parts, labels and capture. The stereo carpentry and framing page covers the cutting and framing rows and not these two. A buyer who wants a mix can combine rows from several pages in one spec.

The site states one published timeline: from a signed spec to a first validated batch, a median of 48 hours where coverage is already deep, and typically three to four weeks where it is thin because operators have to be staffed. The site does not say which applies to either row and promises nothing faster, so ask when you request the quote.

Collection is on demand, so a focused brief is a normal starting point. One piece of hardware fitted from mark to final screw is a reasonable first unit, and a Spec Pilot is a fast way to prove the fill rate and reject rate before you scale.

Data is delivered in RLDS, LeRobot, WebDataset, HDF5, zarr or Rerun .rrd, or as raw streams with metadata, with the episode schema as the canonical layout. QA runs once, against the canonical episode, before any format is cut. The formats table sets the site’s delivery next to each maintainer’s own documentation, linked under Sources.

The license is perpetual, irrevocable and worldwide, buyer-owned and non-exclusive by default, with exclusivity available as a separate term and the executed order form governing.

Firsthand captures humans, not robots. Episodes carry human hand pose and human action labels, so using them for a specific robot means your own retargeting step. The site does not state robot actions, teleoperation data or any claim about a particular model or embodiment for this task.

What the data gives a lab is the human side of close-range fitting: the order of steps, how the hands hold a part and a tool, and where a step begins and ends. A lab can use that for imitation, for segmenting a task into steps or for evaluation. Each of those is the buyer’s own use of the data.

Start with the free sample pack to check the loader and label quality, then move to a Spec Pilot: a small, fast first batch against the written spec that proves the fill rate and reject rate on your conditions before you scale.

A good cabinet hardware brief names the setting, the parts, the stages of the job, how each stage starts and ends, what counts as a reject and the privacy rule for the workshop. The site’s skill-spec guide walks through that structure. Then request a custom dataset.