Egocentric Quilting and Needlepoint Data

It is first-person video of one person doing textile hand work, recorded from that person’s own point of view. A paid, consented contributor wears the capture rig and performs the task written in a skill spec, and each attempt is delivered as an episode with synchronized streams, hand pose and labeled action segments. It is human capture. It is not robot or teleoperation data, and it is not a fixed corpus you download.

People reach the topic through quilting robot training data, egocentric needlepoint video, a fabric piecing and counted stitch dataset or custom textile arts data collection. They describe one request from different angles: write the task down, collect it on demand, and receive validated episodes. Because the data comes from people and not from a robot, using it for a particular robot still needs your own retargeting step.

Two, listed in lib/commercial-catalog-data.ts under the Textile Arts & Stitching Studio environment in the Creative Workshops & Crafts sector. The environment is described with examples of hand embroidery, tailoring studio, quilting, needlepoint, custom garment fitting and hemming. The row "Quilting" reads: "Cut and piece fabric blocks, layer with batting and quilt the layers together." The row "Needlepoint" reads: "Stitch yarn through canvas in a counted pattern with a tapestry needle." Both are rated Medium, both list the capture setup "Phone (head-mounted) / multiview rig", and both are Ready to deploy.

The site defines Ready to deploy as contributors and site access being set up, with capture starting once a spec is signed. The other seven rows of the studio are not the subject here: "Custom garment fitting", "Hand embroidery", "Hemming" and "Tailoring" are also Ready to deploy, and "Crochet", "Embroidery" and "Knitting" are Available on request. This page names only the tasks and items in the two rows above. It states no delivered hours, episodes or footage for either row.

The egocentric embroidery and hand stitching page covers hand-stitching rows, including "Needlepoint", but it states that quilting is outside its scope. This page takes quilting and needlepoint together as its main subject. The egocentric knitting and crochet page covers the two Available on request rows, "Knitting" and "Crochet". The egocentric sewing and textile handling page spans cutting, pinning, hemming, stitching and handling fabric across several environments and reproduces the studio rows as one of its settings.

The stereo embroidery and tailoring page covers all nine studio rows as stereo capture and MCAP delivery, and it states that stereo applies through the site’s general stereo offer and not through these rows. This page is first-person capture of the two rows as the catalog lists them, so it is not a stereo page. The first table below reproduces the two rows, and the second sets the pages side by side. The site has no separate egocentric tailoring and alterations page, so none is compared.

The catalog lists "Phone (head-mounted) / multiview rig" for both rows. The site describes that setup for the studio rows as a head-mounted phone recording the first-person view of real work on site, with a multiview rig adding synchronized third-person angles when the spec asks for them. The catalog does not say which of the two applies to each row, so the spec names the one you want and a Spec Pilot confirms it.

The site does not state a stereo pair for these rows, and this page does not describe the capture as stereo. It names no studio, school, workshop, city or region, and it lists capture at a customer’s own facility as not offered, so this page claims neither. Camera count, resolution, frame rate and lighting for these rows are Not stated. Capture is run to your skill spec using the settings and contributors Firsthand arranges.

The row names three jobs: cutting and piecing fabric blocks, layering with batting, and quilting the layers together. Each is a natural cut. A spec that says "make a quilt" gives one very long action segment. A spec that names cutting, piecing, layering and quilting gives a sequence you can slice by step, with each action segment carrying start_ns, end_ns, verb and noun.

Which verbs and nouns are used, whether a step is a single seam or a whole block, and whether one episode is one step, one block or a whole piece are spec decisions. The site does not set an episode length or a step count, states no block size or quilt size, and names no cutting tool, sewing machine or hoop. How "quilt the layers together" is done, by hand or by machine, is not stated in the row, so a spec that cares should say.

Contact events, which the site describes as the moment a hand meets an item, can mark when a finger or a tool touches the fabric. The site does not say whether a layer of batting or a fold of cloth counts as the item, so the spec settles what a contact means for your task.

The row describes yarn stitched through canvas in a counted pattern with a tapestry needle. The counting is the part a robot cannot read from a single frame, because the same motion repeats while the position moves across the canvas. A spec should say whether an episode is one stitch, one row of stitches, one motif or the whole piece, and what a finished step is.

The site gives no pattern, chart, canvas count, stitch type, yarn weight, stitch count or finished-piece label for needlepoint, and it lists no label for whether a count was correct. Action segments mark when each step starts and ends, and hand pose carries how the fingers hold the needle and the canvas. Anything beyond that, such as a position-in-pattern annotation, is Not stated and would be agreed in the spec.

At a stitching bench the hands sit between the camera and the work, and a needle tip or a length of yarn can vanish behind a finger, a fold of fabric or the canvas edge. 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 a 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 tapestry needle or a strand of yarn counts as such a part is Not stated, and so is a mask accuracy figure for yarn or needles. A spec that needs a clear view of the needle entering the canvas should name that step, ask for a multiview angle, and confirm it on a Spec Pilot.

Fabric blocks, batting, canvas and yarn 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 a single block, the layered stack, the canvas or the needle, 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. Layers that stack and fold, or a canvas that sits under the hand, can change what is in view. The site states no behaviour for particular fabrics, batting or yarn, so how they affect depth or masks is Not stated.

The site states ten hardware-synchronized streams under 2 ms, including head and wrist cameras, depth with confidence maps, IMU, 3D hand and body pose and instance segmentation, plus action segments and contact events. Hand pose is delivered for both hands with a per-joint visibility flag. The site lists the standard label fields as hand_joints, hand_visible, contact_events, objects and instance_masks.

Which of these streams apply to the head-mounted phone or multiview setup the two rows list is Not stated, and is set in the skill spec. The site lists no label for the quilting or the stitching itself: no stitch type or count, seam or pattern quality, block alignment, completion score, contact force or thread tension reading. It provides one task instruction per episode and no other language annotation for these rows.

Every contributor signs a written release before capture, and each delivered episode links to that record through consent artefact IDs in the batch data card. The consent chain has six checkpoints, and a break at any one discards the footage rather than shipping it at a lower grade. At ingest, automated detection finds faces, plates and screens and blurs them irreversibly, and speech is scrubbed of names, addresses and card numbers while ambient audio is kept.

A human reviewer confirms that no un-consented person is identifiable in any frame, and footage that cannot be de-identified is discarded. An undocumented bystander is also a machine-checked reject criterion. The anonymization record is delivered with the data. Neither row involves a client, but a spec that would put a second person in frame should settle consent for that person first. 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. This is not legal advice, and the executed agreement controls.

Fabric types, batting types, yarn types and weights, canvas types and counts, needle sizes, patterns and charts, block and quilt sizes, stitch counts, machine and tool brands, cycle times, error rates, item counts and what a finished piece means are Not stated. They are set in the written spec, and this page claims none of them. Nothing here reports a robot result, a benchmark or a model improvement.

No dataset in the off-the-shelf catalog names quilting or needlepoint, so for these two tasks this is custom collection and not an off-the-shelf purchase. A few listed datasets name sewing, stitching and knitting among other tasks, and they are the place to look for adjacent coverage. The off-the-shelf catalog page lists the current datasets and figures.

Custom collection runs 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 studio rows, so the timeline is quoted against your spec. Collections are billed per validated hour, and rejected episodes and recaptures are never billed.

The site states program milestones beside the per-batch rule: a pilot reaches its first delivery in two to three weeks, and a standing program reaches full rate in four to six weeks. Both assume coverage is in place. After the first batch, on-demand top-ups run against the same accept criteria, so you can add steps or a second row without rewriting the spec. The site does not state a turnaround for those later batches, so plan them in your quote.

Speed comes from a warm pipeline, not from cutting consent, review or anonymization. The site publishes no price list, minimum order or batch size for this task family, so those are set in your quote. A focused brief, such as one row and one step, is a normal starting point.

Delivery is in the format your stack already reads: RLDS, LeRobot, WebDataset, HDF5, zarr or Rerun .rrd, or raw streams with metadata. Readable converter source ships alongside the data so the schema can be retargeted to your own loader. Firsthand also offers MCAP delivery. The official documentation for three of these formats is linked in the sources below, and the MCAP hub page and glossary cover the container.

The site states no topic names, message schemas or rates for an MCAP file for these rows, so none are given here. Name the streams that matter in the skill spec and confirm them on a Spec Pilot batch. The license is perpetual, irrevocable and worldwide, buyer-owned and non-exclusive by default, with exclusivity available as a separate term.

Describe the task and where it starts and ends: cutting and piecing blocks, layering with batting, quilting the layers together, or stitching yarn through canvas in a counted pattern. Say whether an episode covers one step, one block, one row of stitches or a whole piece, what counts as the object and as a finished step, whether the work is by hand, the setting and lighting, the privacy rule if a second person might appear, and name the variation your pipeline is missing.

Then name the streams you want and the delivery format. 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.