Egocentric Commercial Janitorial Data

Egocentric commercial janitorial data is real-world, first-person video of a person doing janitorial and facilities cleaning in offices and commercial buildings. The site states this directly. Under the Commercial Cleaning & Janitorial environment, in the Facilities & Cleaning sector, a catalog row named "Commercial Cleaning & Janitorial" reads "Janitorial and facilities cleaning across offices and commercial buildings." It is recorded from the worker viewpoint, with depth, 3D hand pose, contact events and frame-accurate action segments aligned on one hardware clock, against a written spec.

People search for this under several names: janitorial robot training data, a commercial cleaning video dataset, egocentric custodial task data or custom facility cleaning data collection. They describe the same thing, a record of how a person carries out the routine work of keeping a commercial space clean, seen from the worker view with both hands and the tool in frame.

This is a wider topic than any single cleaning chore. The site already has pages for floor cleaning and mopping, wiping and surface cleaning, bathroom cleaning, window cleaning, hotel room turnover and trash and recycling. This page covers the commercial setting as a whole, the environment where those chores are done as janitorial work, and it points to those pages for single-chore detail rather than repeating it.

This page states only what the site already states. Where a detail is not in the catalog or the delivery terms, it is marked Not stated and left for the written spec. It makes no hygiene, sanitation, disinfection, chemical, safety or certification claim.

The Commercial Cleaning & Janitorial environment holds 15 catalog rows. Its examples line reads "Office building, shopping mall, airport, industrial facility -- mopping, vacuuming, surface sanitizing, restroom cleaning." Eight of the rows are rated Easy and seven are rated Medium, and none is rated Hard. Eight are listed as Active collection and seven as Ready to deploy.

Three rows speak most directly to janitorial work. The general row, "Commercial Cleaning & Janitorial," reads "Janitorial and facilities cleaning across offices and commercial buildings," and is rated Easy and Ready to deploy. Vacuuming reads "Vacuum carpets and hard floors in offices and corridors, moving chairs as needed," and is rated Easy and Active collection. Restroom Restock & Disinfection reads "Refill dispensers and disinfect all restroom fixtures and touch points," and is rated Easy and Ready to deploy.

Around them sit the floor, carpet, restroom, surface and window rows: Mopping, Floor Stripping & Waxing, Carpet Extraction, Surface sanitizing, Restroom cleaning, Event Restroom Sanitation, two window rows and Drain Clearing & Cleanup. Some chores appear as two rows with different wording. The table below lists all fifteen in the catalog words. A buyer who needs only one task can ask for it alone, and one who needs a fuller round across several can ask for them together. The catalog does not define a round, a route or an order of tasks, so any sequence is a spec choice and not a stated task list.

The site does not state which building types a given row was recorded in, which cleaning products or equipment models are used, how long a task takes, how many sites a collection uses or what a finished space looks like. Those are Not stated and are set in the written spec. The catalog also does not name a bin-emptying or janitorial-cart row, so this page does not describe one.

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 events on grasp and release, and frame-accurate action segments. For janitorial work, hand pose tracks the grip on a mop handle, a vacuum or a squeegee pole, contact events mark when the hands meet a tool, a dispenser or a surface, and action segments mark each step of a task.

The site also states 200 Hz 6-DoF IMUs on the Firsthand capture rig. All 15 rows in this environment list the capture mode Phone (head-mounted) / multiview rig, which the site describes as a head-mounted phone recording the first-person view, with a multiview rig adding synchronized third-person angles when the spec asks for them. Which streams a given janitorial collection includes is set in the spec, so confirm the stream list before you plan a training pipeline around a specific sensor.

Object labels use the terms your spec sets. Building types, cleaning products, equipment models, and a task-completion or cleanliness label are Not stated. If your policy needs a label such as "floor done" or "restroom restocked," define it in the spec and the accept criteria. This page makes no head or camera pose, mapping or SLAM claim as a deliverable.

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. Janitorial episodes use the same schema as every other Firsthand collection, so a loader that reads another Firsthand dataset reads this one without changes.

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. Billing is per validated hour, and every collection ships under a perpetual, buyer-owned commercial license.

The site does not state a number of episodes, a batch size or a total hour count for any janitorial row, so those are quoted against your spec and not assumed here.

Janitorial work is repetitive hand-and-tool work in shared commercial spaces. A person pushes a vacuum through a corridor, moves a chair, wrings a mop, refills a dispenser or wipes a desk. A policy meant to learn a sequence like that needs to see how the hands hold and move each tool and when they touch the surface, not only the finished state of the room.

First-person video suits that need because the view follows the worker, so the tool, the surface and both hands stay in frame. 3D hand pose and contact events add the grip-level detail that plain video leaves implicit, and action segments mark where each task starts and stops so a model can learn the order of a round as well as the motion within each step.

This page quotes no robot result, no recovery rate, no throughput figure and no benchmark. It does not claim that any model has been trained on this data, and it makes no hygiene, sanitation, disinfection, chemical, safety or certification claim. The site states a set of catalog tasks and a capture and delivery method, and the value to a given policy is something a pilot can test against your own accept criteria.

For the capture mode these rows use, the site describes a head-mounted phone that records the first-person view of real work on site, with a multiview rig adding synchronized third-person angles when the spec asks for them. That is the only capture mode this page states for janitorial work.

The site does not state whose building that is, how many sites a collection uses, or which regions or cities are covered. Elsewhere it states that Firsthand does not capture at a customer own facility, so this page makes no claim of capture there. Environment, access and schedule are agreed in the spec.

Every episode links to a signed contributor release through a documented consent chain, and the consent record is part of the delivery. The site states that faces, plates and screens are detected and irreversibly blurred at ingest, and that a reviewer confirms the result before anything ships. The anonymization record is delivered with the data.

That is the whole of the privacy wording the site states. It does not state a rule for data stored on a device, so none is claimed here, and you should raise it in the spec if it matters to you.

The catalog lists eight janitorial rows as Active collection and seven as Ready to deploy. The site defines Active collection as recording in this environment running now, so a spec can join an existing capture stream. It defines Ready to deploy as contributors and site access being set up, with capture starting once a spec is signed. Neither status is a delivery date, and the timeline is quoted against your spec before any capture begins.

The site publishes two per-batch speed figures. 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 which of the two applies to janitorial work, so the quote says.

On-demand top-ups run against the same spec and accept criteria after each training run. The site does not state a batch size or a later-batch turnaround for any janitorial row, so those are quoted against your spec.

You describe the janitorial tasks and the round 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, rejected items are recaptured at our cost, and every collection ships under a perpetual, buyer-owned commercial license.

Useful things to settle in the spec are the tasks you want covered and their order, the building type, the equipment, any label you want for task completion, and the formats you want delivered. Each of those is Not stated by the site today, which is why the spec is where they are written down.