Stereo Embroidery and Tailoring Data

It is first-person video of a person doing studio textile work, recorded with two calibrated cameras, with hand pose and sensor streams alongside. People reach it through several phrasings: embroidery robot training data, stereo hand stitching video, an egocentric tailoring and hemming dataset, or custom textile arts data collection at scale. 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, so 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, so using it for a specific robot still needs your own retargeting step. Nothing on this page is safety guidance. Firsthand is not affiliated with Foxglove or the MCAP maintainers.

Nine, listed in lib/commercial-catalog-data.ts under the Textile Arts & Stitching Studio environment in the Creative Workshops & Crafts sector, whose examples line names hand embroidery, tailoring studio, quilting, needlepoint, custom garment fitting and hemming. "Hand embroidery" is described as stitching decorative patterns into fabric by hand with needle and embroidery thread. "Needlepoint" is described as stitching yarn through canvas in a counted pattern with a tapestry needle. "Hemming" is described as measuring, pinning, pressing and stitching hems on trousers and skirts by hand or machine. "Tailoring" is described as cutting, fitting and sewing garments to measure with machine and hand finishing. "Custom garment fitting" is described as pinning and marking a garment on the client and adjusting seams and length for fit. "Quilting" is described as cutting and piecing fabric blocks, layering with batting and quilting the layers together.

The three remaining rows are "Embroidery", described as machine and hand embroidery production of garments and textile goods, "Knitting", described as hand and machine knitting of garments and textile products, and "Crochet", described as producing crocheted garments and textile products to commercial order. Custom garment fitting and Tailoring are rated Hard and the other seven are Medium. Six rows are Ready to deploy, which the site defines as contributors and site access being set up, with capture starting once a spec is signed. Embroidery, Knitting and Crochet are Available on request, which the site defines as work that is scoped and recruited to spec when a buyer asks. The site states no delivered hours and no delivered stereo footage for any of the nine rows, so this page claims none.

This page is the stereo and MCAP treatment of the nine studio rows. The egocentric embroidery and hand stitching page covers the hand-stitching slice of the cluster, including Hand embroidery, Needlepoint, Embroidery, Hemming and Tailoring, as first-person video, and it also reproduces two hand-stitching rows from other environments. The egocentric knitting and crochet page covers the Knitting and Crochet rows the same way. Neither has stereo as its subject, and they remain the pages for the first-person framing and the row-by-row detail. The site has no separate egocentric tailoring and alterations page, so none is compared here.

The stereo garment sewing and fabric cutting page covers four Textile & Apparel Production rows, all Available on request, aimed at industrial sewing machines and cutting tables. The stereo garment finishing and pressing page covers one row, "Finishing and pressing", rated Easy and Available on request. This page is stereo capture of studio embroidery, hand stitching, hemming and tailoring tasks. The first table below reproduces the nine rows, and the second sets the pages side by side.

The catalog lists "Phone (head-mounted) / multiview rig" for all nine rows. The site does not state a stereo pair for any of them. Stereo applies 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 catalog does not say which of its two modes applies to each row, and the site does not state capture at a customer’s own studio, so the setting is agreed in the spec.

The site describes two stereo setups and does not say which one an order for these rows 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. Camera count, resolution and frame rate for the textile rows are Not stated.

Hand embroidery and needlepoint are close work: the needle sits in the fingers for the whole task, one hand guides the cloth or canvas and the other works the thread. The catalog itself names "needle and embroidery thread" and "a tapestry needle", and the tool, the material and both hands occupy a few centimetres of the frame at once. A stereo pair recovers depth from the calibrated frame pair rather than from a standalone depth sensor, so the geometry of the cloth and the hands sits in the same capture that shows the fingers working.

A stereo pair adds depth, not a verdict on whether a stitch is well made. The site lists no stitch-quality, seam or completion label for these rows, and none is claimed.

Each row gives a natural cut. Hand embroidery stitches a decorative pattern. Needlepoint stitches yarn through canvas in a counted pattern. Hemming measures, pins, presses and then stitches. Tailoring cuts, fits and then sews, with machine and hand finishing. Custom garment fitting pins and marks, then adjusts seams and length. Quilting cuts and pieces blocks, layers them with batting and then quilts the layers together.

A spec that says "hem the trousers" gives one long action segment. A spec that names measuring, pinning, pressing and stitching gives a sequence you can slice by step. Whether each of those is its own segment, and whether one stitch, one seam or a whole garment is one episode, is a spec decision, not a default. The site does not set an episode length or a step count. Contact events can mark when a finger or a tool meets the cloth, the canvas or the garment, where the spec asks for them.

Hemming and Tailoring are described as hand or machine and machine with hand finishing, and Embroidery is machine and hand production. A spec that wants only the hand part should say so, and one that wants a machine pass should name it. The site states no sewing machine, iron or press brand and no machine speed for these rows, and how a spinning needle bar or a moving foot is handled by a stereo pair or the depth stream is Not stated.

At a stitching bench the hands sit between the camera and the work, and a needle tip or a length of thread can vanish behind a finger or a fold of cloth. The site states that the hands occlude the work and that fingers disappear behind objects. That is why each hand joint carries a visibility flag, why contact events mark when the hand or a tool meets a surface, and why instance masks are recorded per frame.

The site says two-handed fine motion and small objects are where occlusion bites, and it states that metric depth on sub-centimetre parts is flagged rather than interpolated. Whether a needle or a strand of thread counts as such a part is Not stated, and so is a mask accuracy figure for thread or needles. A spec that needs a clear view of the needle entry, the stitch or the seam should name that step, ask for wrist views or a multiview angle, and confirm it on a Spec Pilot.

Cloth, canvas, batting and thread have no fixed shape, and the site states that masks for deformable objects are shipped per frame rather than propagated. Instance masks are stored at the depth resolution of 848×480 so they line up with the depth stream. What counts as the object, whether the fabric, the garment, the needle or the thread, is a spec decision.

That decision matters because of two machine-checked rules. An episode is rejected if the manipulated object is out of frame for more than 15 percent of it, and if more than 35 percent of depth pixels in the manipulation region are invalid. The site also states that depth degrades on wet and reflective surfaces. It states no behaviour for sequins, metallic thread, pale or dark cloth or patterned fabric, so how those affect depth is Not stated. If your policy depends on a material, say so in the spec and settle it on a Spec Pilot.

The Custom garment fitting row pins and marks a garment on the client, so a fitting puts a second person in the frame. The site states a consent chain for contributors, not a separate process for a client being fitted, so how a client’s consent is handled is Not stated and has to be settled in the spec before capture. An undocumented bystander is a stated reject criterion.

For every delivered episode the site states that faces, plates and screens are detected and irreversibly blurred at ingest, that a reviewer confirms the result before anything ships, and that footage that cannot be de-identified is discarded. The anonymization record ships with the data, and every episode links to a signed contributor release through a documented consent chain. The site names no studio, school, tailor, atelier, city or region and lists capture at a customer’s own facility as not offered, so this page claims neither. Tool safety is Not stated: the site gives no contributor safety protocol, injury-prevention guidance or certification for any textile row, so this page describes none and gives no safety instructions. This is not legal advice, and the executed agreement controls.

Fabric types, thread types and counts, needle sizes, hoops and frames, stitch types and counts, pattern designs, garment sizes, machine and tool brands, cycle times, error rates and what a correct or finished piece means are Not stated. They are set in the written spec, and this page claims none of them.

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 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 these rows the site states hand pose, a stereo-derived depth stream with confidence maps, IMU, contact events, action segments, per-frame instance masks and one task instruction per episode 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 by the needle and the cloth is the first risk, under the visibility flag above. Two-handed fine motion with thin needles and thread is the second, and how often it leaves the frame is Not stated. Calibration is measured at the start and end of every session under a 0.28 px RMS gate, and an episode is rejected for a sync breach beyond the ceiling for more than three consecutive frames, for the manipulated object out of frame for more than 15 percent of the episode, for an undocumented bystander or for a mismatch against the named skill spec.

A reject log ships with every batch, and rejected items are recaptured at the provider’s cost. The fifth table below lists the failure modes beside what the site states.

The October 2026 off-the-shelf catalog lists ~324,800 hours across 18 datasets, of which stereo is ~64,800 hours across 9 datasets. Two stereo datasets name nearby garment work: S2 (~5,000 hours; factories, hotels, kitchens and cafés, with sewing among the listed tasks; stereo RGB, task labels and hand keypoints) and S4 (2,984 hours; yarn mill, textile and garment, with stitching and knitting among the listed tasks; head-mounted stereo with synced left and right RGB, IMU and audio). Neither names embroidery, needlepoint, hemming, tailoring, quilting or garment fitting, so for those tasks this is custom collection, not an off-the-shelf purchase.

It is 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 applies to the textile rows, so the timeline is quoted against your spec. Collections are billed per validated hour, and rejected episodes and recaptures are never billed.

Describe the tasks and where each starts and ends: stitching a pattern by hand, counting stitches across canvas, measuring, pinning, pressing and stitching a hem, cutting, fitting and sewing a garment, pinning and marking a fitting, or piecing and quilting. Say whether an episode covers one stitch, one seam or a whole piece, what counts as the object and what counts as a finished step, the setting and lighting, the privacy rule for the studio and for any client in frame, and name 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 sixth table below shows what each spec element decides.