Stereo Kitchen Deep Clean and Degreasing Data

It is first-person video of a person degreasing and deep cleaning a commercial kitchen, recorded from their own point of view with two calibrated cameras and with hand pose and sensor streams alongside. The tasks are the two rows the site’s catalog lists in Restaurant & Commercial Kitchen: spraying degreaser on hoods, walls and counters, scrubbing with pads and rinsing, and emptying and scrubbing fryers, hoods, floors and shelving with degreaser, brushes and mops. Stereo describes the capture, and MCAP describes the container. They are separate choices.

The site does not state that stereo is already running on these rows. The catalog lists a head-mounted phone or a multiview rig as the capture setup. Stereo applies only through the site’s general stereo offer: a stereo pair is a capture choice agreed in the skill spec and confirmed on a Spec Pilot. The data is human capture by paid, consented contributors, not robot or teleoperation data. Firsthand is not affiliated with Foxglove or the MCAP maintainers.

Two, both in Restaurant & Commercial Kitchen under Food & Beverage, in the catalog data file lib/commercial-catalog-data.ts. The catalog lists examples for the environment such as a sit-down restaurant kitchen, a café, a coffee shop, an artisan bakery and a patisserie. "Degrease and scrub kitchen surfaces using degreaser and scrub pads" is described as spraying degreaser on hoods, walls and counters, scrubbing with pads and rinsing the surfaces clean, is rated Easy and is listed as Active collection.

"Restaurant Kitchen Deep Clean" is described as emptying and scrubbing fryers, hoods, floors and shelving with degreaser, brushes and mops, is rated Medium and is listed as Ready to deploy. Both list a head-mounted phone or a multiview rig as the capture setup. The site says nothing further about what each status means in volume, hours or delivered stereo footage, so this page claims none, and what applies to a given order is confirmed when you request the quote. This page names only the tools and surfaces that appear in the two rows. The first table below reproduces them as the catalog states them.

"Trash & Recycling Removal in Kitchen" sits in the same environment. It is described as pulling full bags and recycling from the kitchen, carrying them to the bins and relining the cans, is rated Easy and is listed as Ready to deploy, with the same head-mounted phone or multiview rig capture setup. It is not a degreasing or deep clean row, so this page treats it as nearby context only and builds nothing on it.

The same applies to the dishwashing and line setup rows in that environment. They are covered by their own pages, listed in the comparison table below, and they are not counted as part of this page’s scope.

The stereo dishwashing and kitchen sanitation page is the closest neighbour. It is stereo and MCAP too, but it centres the dish pit and treats the two rows on this page as half of a four-row set. This page narrows to the degreasing and deep clean rows alone, and goes further into the surfaces, tools and failure modes that belong to them. The stereo food plant washdown page is stereo and MCAP for the two plant rows, and neither of the kitchen rows appears there.

The egocentric food plant sanitation and washdown page lists both kitchen rows beside the two plant rows, with an egocentric catalog treatment and no MCAP treatment. The egocentric wiping page covers wipe-and-reset rows, not these. The egocentric kitchen line setup page covers the two pan and line rows. The comparison table below sets the pages side by side.

Deep clean work is about where the hands, the tool and the surface sit relative to one another: how far a spray bottle is from a hood, where a scrub pad has already been on a wall, how a brush reaches into a shelf. A stereo pair recovers depth from the calibrated frame pair rather than from a standalone depth sensor, so the tool, the surface and the hands share a frame.

The site gives no depth accuracy figure and no label for a degreased surface, a clean fryer or a finished shelf, so whether those are resolvable in depth is Not stated. The site says that wet and reflective surfaces, such as a rinsed pan, a wet countertop and a steel sink, degrade active-stereo depth. How a stereo pair behaves on them is Not stated, and a kitchen spec should expect the same care. Capture at your own facility is listed as not offered.

The catalog states one capture setup for both rows: a head-mounted phone or a multiview rig. The site describes two stereo setups and does not say which one a kitchen order uses. The hard-environments stack is a global-shutter 4K/60 head camera with a 120° lens, two 1080p/60 wrist cameras, a depth sensor at 848×480 and 30 fps, and 200 Hz 6-DoF IMUs. The off-the-shelf stereo head rig, dataset 06, is dual fisheye cameras at 1080p per eye, a 62 mm baseline, per-unit calibration, a 9-axis IMU and hand-tracking output.

Which rig, resolution and streams a kitchen deep clean order uses is Not stated on the site, and the third table below sets what is stated beside what is not. Intrinsics and extrinsics are re-solved at the start and end of every session, and streams are hardware-triggered onto one clock with a 2.0 ms cross-stream ceiling. The catalog states that a multiview rig adds synchronized third-person angles when the spec asks for them, and wrist cameras give a closer view of the hands. The site does not state how capture hardware is protected from water or spray.

MCAP is an open source container file format for multimodal log data, supporting multiple channels of timestamped pre-serialized data. It is serialization-agnostic, so messages can be Protobuf, DDS (CDR), ROS, JSON and more. Its design is row-oriented and append-only, it stores schemas alongside the data, and it offers an optional index plus optional LZ4 or Zstandard chunk compression.

Reader and writer libraries exist for C++, Go, Python, Rust, Swift and TypeScript, and the mcap CLI can inspect, merge and split files. The foxglove/mcap repository is MIT-licensed; that is the MCAP project’s licence, not the licence of a Firsthand delivery. The MCAP hub page and the glossary entry cover the format in full.

Firsthand offers MCAP delivery, and dataset 06 is stated as shipping as MCAP. For kitchen work the site states hand pose, a stereo-derived depth stream with confidence maps, IMU, contact events, action segments and per-frame instance masks elsewhere. What it does not state is which of those are channels in an MCAP file, how they are named, what encoding they use or at what rate.

No topic name, message schema or rate is stated on this page, and none should be assumed. Name the streams that matter in the skill spec and confirm them on a Spec Pilot batch. The fourth table below sets each stream beside what the site states about it.

Hand occlusion is the first risk. Fingers go behind a scrub pad, a brush or a mop head, and a tool can hide the surface it is working. The visibility flag on each joint is the answer the site gives, and contact events mark when a hand meets an item. A spec can also ask for wrist views, so a grip the head camera loses is still recorded.

Wet and reflective surfaces are the second. The site ships per-frame depth confidence maps for kitchen captures, so low-confidence regions can be masked rather than trusted, and per-frame instance masks for wet or reflective cookware, because masks propagated frame to frame do not track them well. It lists no degreased, clean or dry label, so what counts as a finished surface is written into the spec up front. Spray, steam, lighting and clutter around a hood or a fryer are not described on the site, so how they affect a stereo pair is Not stated.

Calibration is the third, because stereo depth is only as good as the rectification behind it: calibration is measured at the start and end of every session under a 0.28 px RMS gate. An episode is also rejected when the manipulated object is out of frame for more than 15 percent of it or an undocumented bystander appears, and a reject log ships with every batch. Force and tactile readings are not among the listed streams.

A deep clean chains steps that look alike on camera: spraying, scrubbing and rinsing a hood, then the same on a wall, then on a counter, or emptying a fryer and then scrubbing it. Action segments carry start, end, verb and noun, so the spec decides where one segment ends and the next begins, and which object is the noun.

The site states no episode length, no segment count and no scrub-stroke label for these rows. Whether an episode is one surface, one appliance or a whole station is set per program. Put that boundary and the verbs you expect into the skill spec, and check them on the Spec Pilot before you scale.

For every delivered episode the site states that faces, plates and screens are blurred irreversibly, and that footage that cannot be de-identified is discarded. Who may be in frame, and how a kitchen’s own staff, menus or labels are handled, is agreed in the spec. The anonymization page has the full statement.

The site states no cleaning-standard, chemical, throughput, accuracy, food-safety or certification claim for these rows, so this page describes none. It also states no restaurant, client or brand name, no degreaser brand or dilution ratio, no equipment model, no surface count and no task time for these rows. Each of those is Not stated.

Off-the-shelf covers what is already in stock: the stereo category lists the stereo datasets and hand-tracking video, and only dataset 06 is stated as MCAP. This page does not claim an off-the-shelf dataset of kitchen deep clean or degreasing, so this is custom collection: custom collection at scale through a network of thousands of capture providers, for a consistent, recurring data flow to labs.

Where coverage is deep, the median time from signed spec to first validated batch is a median of 48 hours; where it is thin, turnaround runs three to four weeks. The site does not say which speed applies to a stereo kitchen order, so ask when you request the quote. Collections are billed per validated hour.

Describe the surfaces and appliances and their order, the degreaser, pads, brushes and mops and their start state, what counts as a finished surface, and the variation your pipeline is missing. Then name the streams you want in the file, the topic naming and message encoding your reader expects, whether calibration ships inside the file or beside it, and whether MCAP is the only delivery or sits beside RLDS, LeRobot or WebDataset.

Most programs begin with a Spec Pilot, a small first batch that proves the spec and accept criteria on real footage before you scale. The fifth table below shows what each spec element decides.