Stereo Part Trimming and Deflashing Data

It is first-person video of a person removing flash from molded parts, recorded from the worker’s own point of view with two calibrated cameras, with hand pose and sensor streams alongside. 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 their 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. Firsthand is not affiliated with Foxglove or the MCAP maintainers.

Two, listed in lib/commercial-catalog-data.ts, both in the Plastic & Light-Assembly Manufacturer environment. "Trimming & Deflashing Parts" is described as "Cut excess plastic from molded parts and deburr edges with hand tools." "Trim and deflash molded parts using flush cutters and deburring blades" is described as "Trim flash from molded parts with flush cutters and clean edges with a deburring blade." Both are rated Easy.

Both are listed as Ready to deploy. The site states no delivered hours and no delivered stereo footage for them, so this page claims none. The environment examples read "Consumer goods assembly, packaging operations, labeling, quality inspection on a production line", which are kinds of work, not named firms, and this page names none. The first table below reproduces the rows as the catalog states them.

This page is the stereo and MCAP treatment of the two deflashing rows. The egocentric part trimming and deflashing page covers the same two rows without stereo as its subject: it goes deeper on the ten synchronized streams, the labels and the steps. The stereo grinding and deburring page covers two Metal Fabrication & Welding Workshop rows that work on weld seams with an angle grinder and files, which is metal, not molded plastic.

The stereo injection molding machine tending page covers two Medium rows in which trimming is one step of a press cycle with heat gloves and part bins. The stereo surface finishing page covers nine sanding, drywall and painting rows, with a mix of statuses. The egocentric plastic chair manufacturing page covers one Medium row for chair shells, legs and hardware. The second table below sets the pages side by side with each status as the site lists it.

The catalog lists "Phone (head-mounted) / multiview rig" for both rows, and the plastic and light-assembly manufacturer environment page lists the same capture setup. The site does not state a head-mounted stereo setup for either row. Stereo applies through the site general stereo offer: a stereo pair is a capture choice agreed in the skill spec and confirmed on a Spec Pilot.

A multiview rig adds synchronized third-person angles, and wrist cameras give a closer view of the hands, when the spec asks for them. Whether a stereo pair is worn, mounted or combined with a multiview rig on these rows is Not stated, and so is where a rig sits relative to the bench. The site also states that Firsthand does not capture at a customer’s own facility, so the bench used is agreed in the spec.

Deflashing is about where a cutter meets the part, how the fingers hold the part and how the blade rides along the edge. A stereo pair recovers depth from the calibrated frame pair rather than from a standalone depth sensor, so the geometry sits in the same capture that shows the hands working. The site gives no depth accuracy figure for these rows, so what a buyer does with that depth is their pipeline’s question.

A stereo pair adds depth, not a verdict. The site states 1080p per eye for dataset 06, and whether that resolves a thin line of flash at working distance is Not stated, so a spec should check it on a Spec Pilot. The egocentric page already supplies the first-person view and the ten streams; stereo is the addition for a buyer who wants the depth and the MCAP file.

The rows give a natural cut: pick up the molded part, trim the flash with flush cutters (or cut excess plastic with hand tools), clean or deburr the edge with a deburring blade and set the part aside. The stages are the catalog’s own wording, except that picking up and setting aside are an illustration of how a spec can cut an episode, not a stated duration.

A spec that says "finish the part" gives one long action segment. A spec that names the steps gives a sequence you can slice by step, and the site does not set an episode length or say how many parts make up an episode. Contact events can mark when a hand meets a part, a cutter or a blade, where the spec asks for them, and the spec decides what counts as the item.

The cutter, the blade and the part can all sit between the lens and the fingers. The site states that fingers disappear behind objects, which is why each hand joint carries a visibility flag, and that reach and grasp frames are where blur and occlusion show up. The egocentric light assembly material says small parts can be hidden by fingers. The head camera is global-shutter, which avoids rolling-shutter skew.

The site gives no figure for how often a finger is hidden on this job, so a spec can ask for wrist views or a second angle and check it on a Spec Pilot. The site states no grip-force, cutting-force or tool-pose stream, and this page does not claim one.

The rows describe cleaning edges with a deburring blade, and a stereo pair can show the edge and the blade in three dimensions. The site lists no flash, burr, edge-quality, defect or completion label for these rows, so none is claimed. Whether an edge counts as clean is a judgement a buyer’s pipeline or spec has to define.

An episode is rejected when the manipulated object is out of frame for more than 15 percent of it, and for a deflashing job the spec decides what that object is: the part, the tool or both. Part material, resin, size, geometry, flash thickness, tolerance, defect rate and cycle time are Not stated.

Cut-resistant gloves, blade hazards and hot parts are Not stated for these rows, and this page makes no safety or compliance claim. The press row elsewhere names heat gloves as a tool, but that is a different row. Tool brands, blade types and cutter models are Not stated.

For every delivered episode the site states that faces, plates and screens are blurred irreversibly, 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, an undocumented bystander is a stated reject criterion, and every episode links to a signed contributor release. This is not legal advice, and the executed agreement controls.

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 and tool occlusion is the first risk, under the visibility flag above. Invalid depth on glossy or low-texture surfaces is the second: an episode is rejected when more than 35 percent of depth pixels in the manipulation region are invalid, and confidence maps ship with depth.

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 an undocumented bystander or for a mismatch against the named skill spec. A reject log ships with every batch. 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. The site names no molded-part trimming or plastics dataset among them, so this work is custom collection, not an off-the-shelf purchase. The off-the-shelf stereo and hand-tracking video can still show the format before you write a spec.

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 speed applies to these 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 steps and where each starts and ends: picking up the part, trimming the flash with flush cutters, cleaning the edge with a deburring blade and setting the part aside. Say whether an episode covers one part, a batch or one step, what counts as the object and what counts as a completed step, 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.