Egocentric Glass and Mosaic Craft Video Data

Egocentric glass and mosaic video data is first-person video of a person doing studio glasswork or tile work, recorded from their own point of view with synchronized sensor streams alongside. The camera sits on the head, close to what a robot’s onboard camera would see at a bench, so the frames show what the hands are doing to the piece rather than a distant view of a studio. The same idea sits behind any glass craft robot training data request or mosaic tiling video dataset, applied here as a custom collection scoped to your tasks.

On this site the data is human capture: a paid, consented contributor wears the rig and performs a task from a written skill spec. It is not robot or teleoperation data. This page differs from the pottery and ceramics page, which covers clay, and from the jewelry and metalsmithing page, which covers metal.

The task catalog lists one relevant environment, Glass & Mosaic Studio, under the Creative Workshops & Crafts sector. It has four rows, and its example line reads "Glassblowing studio, stained glass, mosaic tiling, glass etching". The first table below lists the four rows as the catalog states them. All four are Available on request, so the work is scoped and recruited to spec when a buyer asks.

Glass cutting is not its own row. Cutting appears inside two rows: the mosaic tiling row says to cut tiles, and the stained glass assembly row says to cut glass pieces to pattern. Scoring, breaking and grinding are not named in any of the four rows, and glass packing is not listed at all. If your program needs one of them, it is written into the skill spec rather than assumed.

Three rows in other environments also name glass or tile work. They are listed in the second table for orientation only; each belongs to its own environment, and this page does not cover them. The site states no tool or equipment models, glass types or thicknesses, studio types or sizes, or cycle times.

Each attempt is delivered as one episode. On this site the standard episode carries action segments, 3D hand pose for both hands with a per-joint visibility flag, contact events that mark when a hand meets an item, an objects list, and per-frame instance masks stored at the depth resolution of 848×480. The capture setup table below lists the stated streams and sync figures.

The catalog lists one capture mode for all four glass and mosaic rows: a head-mounted phone with a multiview rig. That is a first-person view of real work, with a multiview rig adding synchronized third-person angles when the spec asks for them. Force and tactile readings are not among the listed streams, and neither is a thermal reading, which matters for the glassblowing row that works with molten glass.

Action segments divide an episode into start- and end-timestamped spans, each carrying start_ns, end_ns, verb and noun. For glass and mosaic work, a segment might be one etch, one tile set or one joint soldered, as written in the skill spec. The site lists no label for the outcome of a step, such as whether a cut was clean.

Validation runs against signed accept criteria written into the spec before capture, and a reject log ships with each batch. Episodes are machine-checked against stated reject criteria, such as a sync breach or the manipulated object out of frame for more than 15 percent of the episode, and a human reviewer then checks each one. Only footage that clears the criteria is billed as a validated hour.

Delivery is in the format your stack already reads: RLDS, LeRobot, WebDataset, HDF5, zarr or Rerun .rrd, or raw streams with metadata, with readable converter source alongside. The license is perpetual, irrevocable and worldwide, buyer-owned and non-exclusive by default, with exclusivity available as a separate term. The executed agreement controls.

The catalog describes the work in steps that are small, sequential and unforgiving: mask the glass and rinse to reveal a design, gather molten glass on the pipe and transfer it to anneal, set pieces in a pattern, fit lead came or copper foil and solder the joints. Pieces are small and are often hidden by the fingers that hold them. That is why hand pose carries a per-joint visibility flag, why contact events mark when the hand meets the item, and why masks are recorded per frame. The challenge table below maps each difficulty to the stream or label this site states.

Clear and reflective pieces are a separate problem. The site lists depth and instance segmentation as streams but states nothing specific to transparent or reflective surfaces, so the table says so rather than implying a fix.

A robot foundation model paper on arXiv organizes its training data as a pyramid, with large quantities of web data and human videos at the base, synthetic data in the middle and real robot data at the top. The same paper says human videos have no explicit action labels, so it learns latent actions to label them. That is robot-model research, not Firsthand data, and Firsthand has no relationship with its authors.

The catalog describes the head-mounted phone as recording the first-person view of real work. The site also states that Firsthand does not capture at a customer’s own facility. Which glass or mosaic studio a given program uses is not stated here and is agreed in the skill spec before capture begins.

Every contributor signs a written release, and each delivered episode links to it through consent artefact IDs. At ingest, automated detection finds faces, plates and screens and blurs them irreversibly, and speech is scrubbed of names, addresses and card numbers. A human reviewer confirms that no un-consented person is identifiable in any frame, and the anonymization record ships with the data.

Whether a contributor’s finished pieces or original designs are blurred or protected beyond the license is not stated. Neither are safety certifications, protective equipment or injury handling for glass work, and this page makes no such claim. Who may be in frame and what the footage may show are agreed in the spec.

From signed spec to first validated batch, the published timeline is a median of 48 hours where coverage is deep and three to four weeks where it is thin and contributors need to be recruited first. All four rows are Available on request, so the range that applies is confirmed when the spec is quoted rather than assumed.

Describe the bench tasks and their order, the pieces and tools involved, what counts as a finished step, and the failure or variation cases your pipeline is missing. That becomes a written skill spec, quoted for reach, timeline and price before any capture begins. Our skill spec guide walks through how to write one.

Most programs begin with a Spec Pilot, a small, fast first batch that proves the spec and the accept criteria on real footage before volume scales. Collections are billed per validated hour. Counts of sites, episodes and hours are set in the skill spec.