Egocentric Video Data for Construction and Skilled Trades

Egocentric construction and trades data is first-person video of real jobsite and skilled-trades work — recorded from a camera positioned near the worker’s own point of view, at a chest mount, head rig, or wrist camera, rather than a fixed overhead or side-angle camera. That viewpoint matters for training a robot or AI system to perform the same task, because it sees the work the way a robot’s own onboard camera would see it, rather than from an angle no robot will ever occupy.

The underlying capture, labeling, QA, and licensing process is the same one used across every custom collection on this site: a written skill spec describing the tasks and setting, paid and consented contributors performing the real work, synchronized sensor streams recorded alongside the video, and per-batch QA before delivery. What changes for construction and trades is the setting and the task list written into that spec, and the fact that this is a thinner-coverage environment on this site — the same relationship this site’s industry page describes between every industry setting it covers.

A good fit for capture is any trade task a crew performs the same way, by design, as part of a defined job: a measuring and marking sequence, a fastening or fitting routine, a finishing pass, a tool hand-off between steps. The kind of structure a written skill spec is built to describe is a named, repeatable task with clear start and end points, not an open-ended description of “construction work” in general.

Highly improvised, one-off, or site-specific tasks that do not repeat in a consistent way are harder to specify cleanly and are usually a worse starting point than a task the trade already performs to a consistent method. If a use case spans tool use and bench-style fitting work rather than a jobsite setting specifically, this site’s tool use and assembly page may be a closer fit — the two pages are complementary rather than overlapping.

A construction or trades collection starts the same way any custom collection on this site does: you describe the jobsite or trade tasks your training pipeline needs covered — which trade, which tasks, which tools, which steps matter, and which failure cases need to be represented. That becomes a written skill spec naming the task list, the setting, the participant mix, capture tier, annotation depth, and accept criteria, quoted for reach, timeline, and price before any capture begins.

Because this is a thinner-coverage environment on this site, the spec and quote also account for the added time it takes to vet and schedule operators with the right site access and trade background, rather than assuming the same availability as a deeper-coverage setting.

Many programs start with a Spec Pilot, a small, fast first batch that proves the spec and accept criteria on a limited set of tasks before scaling to full volume. That lets a buyer confirm coverage and label quality on a single task or trade before committing to capture across a larger program.

The same capture tiers used across every Firsthand collection apply to construction and trades work: a phone in a chest mount for a fast, low-friction capture, up to a head rig or multi-camera setup for richer stream coverage, chosen to match the spec’s requirements. Streams recorded alongside the video can include hand pose and action segments, synchronized to the video under the same hardware-alignment standard used across the catalog, with the exact stream set specced per program.

A jobsite presents constraints a controlled workspace does not: other trades and equipment sharing the same space, site schedules that leave limited capture windows, and conditions that vary site to site rather than one fixed layout. Capture is planned around the site’s own schedule rather than requiring the jobsite to pause for the camera.

Raw video alone does not tell a model what happened in a task, which is why captured footage is broken into the same structure used across every collection on this site: action segments marking the start and end of each discrete step, episodes grouping a full task attempt from start to finish, and hand pose tracking the fine manipulation that a video frame alone does not resolve.

A fastening or fitting task, for example, segments into its component steps rather than being labeled as one undifferentiated clip. That step-level structure is what lets a training pipeline learn the individual actions inside a task, not just recognize that “a task happened.”

Failure and recovery cases are part of that same labeled structure — a misaligned fastener, a redo of a measuring step, a dropped tool and recovery. Representing how a real worker catches and corrects a mistake, not only the clean version of a task, is part of what a written spec can call out as a requirement for the batch.

Every custom collection ships under a documented consent chain, with each episode linked to a signed contributor release. Faces and screens anywhere in frame are detected and irreversibly blurred at ingest, with a reviewer confirming the result before anything is delivered.

On a shared jobsite, that same anonymization standard applies to anyone who appears in frame who is not a consenting contributor to the collection — the same bystander-protection approach this site applies in every other setting where people other than the task performer are present. A clip where someone in frame cannot be properly de-identified is handled the same way it is everywhere else on this site: discarded rather than shipped with a caveat.

The buyer. Every custom collection, including a construction and trades program, is delivered under a perpetual, non-exclusive commercial license by default, with exclusivity available, so the finished dataset can be trained on and productized without a research-only or unstated licensing arrangement.

Delivery is billed and licensed the same way as any other collection on this site — against the accept criteria named in the written spec, with rejected items recaptured at no cost to the buyer before anything ships. Pricing is quoted per spec rather than published as a flat rate, because the right capture tier, annotation depth, and volume vary by program; see the pricing page for how a quote is built.

Construction and trades data ships in the same formats available for any custom collection on this site: RLDS, LeRobot, and WebDataset, plus HDF5, zarr, and Rerun .rrd, or raw streams with metadata if your pipeline needs the unprocessed form. Readable converter source ships alongside the data, so the schema can be retargeted to your own loader rather than locking you into one format.

Describe the jobsite or trade tasks, tools, and failure cases your training pipeline is missing. That becomes a written spec covering the task list, setting, participant mix, capture tier, annotation depth, and accept criteria, quoted for reach, timeline, and price before capture begins.

From there the program runs through the same end-to-end pipeline as every Firsthand collection: contributor recruiting, capture on site, per-batch QA, anonymization, and licensed delivery, with on-demand top-ups available afterward against the same accept criteria. Because this is a thinner-coverage setting, a first batch typically lands in three to four weeks rather than the faster median available in deeper-coverage settings — a timeline that is quoted up front rather than assumed from a different industry.