Now recruiting capture teams worldwide. Get paid to record first-person video of real work.

Apply to capture

Task family

Egocentric Table Saw and Wood Cutting Data

Measuring and marking a panel, cutting it on the table saw, shaping a curve with a jigsaw, dimensioning and cutting timber: workshop cutting is a chain of tool-led steps where the line, the tool and both hands all have to stay in view. This page covers egocentric table saw data for the two catalog rows that name this work, as first-person wood cutting and jigsaw video: what the capture is, how the rows are rated and labeled, and what the site does not state. We run custom woodworking data collection fast and on demand against a spec you write, as a panel cutting robot dataset, at scale through a network of thousands of capture providers.
Catalog rows
Cut and shape wood panels using a table saw and jigsaw; Wood Cutting
Skill and status
Both Hard, Active collection, as the catalog lists them
Catalog capture setup
Phone (head-mounted) / multiview rig
License
Perpetual, buyer-owned

Why egocentric here

What does first-person capture add to wood cutting video?

01

The panel, the line and the tool in one frame with the hands

A table saw cut and a jigsaw curve are each defined by a line on a surface and a tool moving along it. A head-mounted camera sits close to what a robot’s onboard camera would see, so the frames show the panel, the line being followed, the tool and the hands on the work rather than a distant view of someone in a workshop.

02

Hands behind work, flagged when hidden

A panel or a tool can hide the fingers that hold it. 3D hand pose carries a per-joint visibility flag, so a hidden finger is flagged as hidden rather than silently guessed, and the two wrist cameras add a closer view of the hands than the head camera alone.

03

Steps in order, labeled as segments

Measuring and marking, cutting on the table saw and shaping with a jigsaw are different steps. Action segments carry start_ns, end_ns, verb and noun as the skill spec defines them, so each step can become its own labeled span instead of one long clip.

04

A spec that names the step and the piece

The site’s skill-spec guide says to name the verb and the noun rather than a broad activity. For this family that means naming the step, such as cut a panel on the table saw or shape a curve with a jigsaw, and where each one starts and ends.

Definition

What is egocentric table saw and wood cutting data?

Egocentric table saw and wood cutting data is first-person video of a person working wood in a workshop: measuring and marking panels, cutting them on the table saw, shaping curves with a jigsaw, and dimensioning and cutting timber. It is recorded from the contributor’s own point of view, with a head-mounted camera and synchronized sensor streams alongside, so the frames show the panel, the line being followed, the tool and the hands together. The tasks are the two rows the site’s catalog lists for this work.

The data is human capture. A paid, consented contributor wears the capture rig and does the task to a written skill spec. It is not robot or teleoperation data, and it is not a fixed corpus you download: the site lists no off-the-shelf table saw or wood cutting dataset, so this work is custom collection. People reach the topic through close cousins such as egocentric table saw data, a panel cutting robot dataset, first-person wood cutting and jigsaw video, or custom woodworking data collection. They describe one request from different angles, and the work is the same: write down the task, collect it on demand, and receive validated episodes.

Catalog rows

Which catalog rows does this page rely on, and what is their status?

Two, both in the Carpentry & Furniture Workshop environment of the Manufacturing & Workshops sector, which the catalog describes as a joinery and furniture workshop with saws, sanders and finishing stations. The first is named "Cut and shape wood panels using a table saw and jigsaw" and reads "Measure and mark panels, cut them on the table saw and shape curves with a jigsaw." The second is named "Wood Cutting" and reads "Dimensioning and cutting timber with commercial saws and cutting equipment."

Both rows are rated Hard, list a head-mounted phone or a multiview rig as the capture setup, and are listed as Active collection. The site says nothing further about what that status means in volume, hours or delivered footage, so this page claims none, and what applies to a given order is confirmed when you request the quote. This page names only the tools and steps that appear in the rows: the table saw, the jigsaw, measuring and marking panels, shaping curves, and dimensioning and cutting timber. The first table below reproduces both rows as the catalog states them.

Hard rows

Why are these two rows rated Hard?

On this site, Hard is the skill level shown in the catalog’s Skill column, one of Easy, Medium or Hard, and this page uses the word in no other sense. Both rows carry it, and so do two other workshop rows, "Woodworking" and "Veneer Lamination & Pressing". The site’s woodworking page calls cutting and shaping the hard rows, and the hard-task hub lists every row rated Hard, grouped by environment.

What the row text itself shows is a chain of steps. The first row packs three jobs into one description: measure and mark panels, cut them on the table saw, and shape curves with a jigsaw. The second dimensions and cuts timber with commercial saws and cutting equipment. Existing site copy adds that woodworking is a sequence of ordered moves with changing tools, that the order matters, and that fingers disappear behind objects in manipulation, which is why hand pose carries a visibility flag. The site publishes no failure rate, tolerance or claim about workers’ skill for these rows, so this page makes none.

Difference

How is this different from the stereo table saw page and the egocentric woodworking page?

The stereo table saw and wood cutting page covers the same two rows as a stereo and MCAP question: what a calibrated stereo pair adds, and what the MCAP container is. This page covers them as an egocentric question: what first-person capture records, which streams and labels the site states, and how the rows are rated. The catalog lists a head-mounted phone or multiview rig on both rows, so the first-person framing here is the one the rows already state, while a stereo pair is a capture choice agreed in the skill spec.

The egocentric woodworking and carpentry page is broader. It treats all eleven rows of the workshop environment, from sanding and finishing to hardware installation, as one family, with these two rows among them. This page narrows to the two cutting rows and gives the table saw, jigsaw and timber cutting steps their own treatment. The comparison table below sets the pages side by side.

Why egocentric

Why capture table saw and jigsaw work from the first-person view?

A cut is a line on a surface and a tool moving along it, with the hands holding and guiding the panel. First-person capture keeps the hands, the tool and the workpiece in frame together, and the two wrist cameras give a closer view of the hands than the head camera alone. The catalog states that a multiview rig adds synchronized third-person angles when the spec asks for them.

A published robot foundation model paper, GR00T N1, linked in the sources below, describes a data pyramid with web data and human video at the base, synthetic data in the middle and real robot trajectories at the top. First-person human video of wood cutting sits in that base layer. Human capture does not replace robot data, and this page does not claim that it does. A second published write-up, GEN-0, also linked below, says "data quality and diversity matters more than sheer volume." For this family that points at the spread of steps and workpieces, which a custom collection can aim at.

Scope

Does Wood Cutting mean a sawmill or a production line?

No claim is made. The "Wood Cutting" row mentions commercial saws and cutting equipment, but the site states no sawmill, industrial production line or mill capability, so this page makes none either. The row sits in a joinery and furniture workshop. Which saws and which equipment a given order covers is Not stated and is agreed in the spec, not assumed. The site also states no saw model, tool brand or workpiece specification for either row.

Capture setup

What capture setup and streams does the site state?

The catalog lists a head-mounted phone or a multiview rig for both rows. The wider Firsthand stack is ten streams: a 4K head camera, two wrist cameras, depth, IMU, 3D hand and body pose, instance segmentation and action labels, hardware-aligned under 2 ms. Every stream is hardware-triggered onto one PTP clock domain, so depth, motion and hand pose line up with the same video frame rather than being approximated afterwards.

The rig is calibrated twice per session, before capture starts and after it ends, and an episode is accepted only if RMS reprojection stays under 0.28 px. IMU data is 6-DoF at 200 Hz from head- and wrist-mounted sensors. Which tier applies to a wood cutting program is set in the skill spec. The site does not state a camera count, resolution or frame rate specific to these rows, and force and tactile readings are not among the streams. The second table below sets what is stated beside what is not.

Labels

How is wood cutting video labeled?

Inside each episode, action segments divide the recording into start- and end-timestamped spans, each carrying start_ns, end_ns, verb and noun. For this family a segment can be one step as written in the skill spec, so the table saw row becomes a run of labeled steps such as measure, mark, cut and shape rather than one long clip. The other standard fields are hand_joints and hand_visible for 3D hand pose, contact_events for the moment the hand meets an object, objects, and instance_masks, stored at the depth resolution of 848×480.

Hand pose is delivered for both hands, 21 joints per hand, with a per-joint visibility flag for occluded fingers. The site does not say whether a panel, a timber piece or a tool counts as the item in a contact event, so the spec settles that. It also lists no label for wood species, panel dimensions, a cut-line deviation or tolerance, or a task-completion score, and no contact force or torque reading. Catalog descriptions describe the task the contributor performs, not a delivered measurement.

Stated limits

What does the site state about the hard parts of recording this work?

Hand occlusion is the first limit the site states: fingers disappear behind objects, which is why pose carries a per-joint visibility flag, and contact events and wrist views help where the head camera loses a grip. Depth is the second. The site says depth degrades on wet and reflective surfaces and handles that with per-frame confidence maps. It says nothing about sawdust in the air or on a surface, or about a moving tool, so those are raised in the spec.

The third is the set of machine-checked reject criteria. An episode is rejected for a sync breach beyond the ceiling for more than three consecutive frames, calibration drift beyond tolerance, the manipulated object out of frame for more than 15 percent of the episode, more than 35 percent invalid depth pixels in the manipulation region, an undocumented bystander, or a mismatch against the named skill spec. A human reviewer then checks each episode against the task spec. A reject log ships with every batch. For wood cutting, the spec decides what the manipulated object is: the panel, the timber or the tool.

QA, consent and privacy

How do QA, consent and anonymization apply?

Validation runs against signed accept criteria written into the spec before capture. Only footage that clears the criteria is billed as a validated hour, and rejected items are recaptured at our cost. 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. At ingest, automated detection finds faces, plates and screens and blurs them irreversibly, and footage that cannot be de-identified is discarded. The anonymization record is delivered with the data.

Workshops can have other people nearby, and an un-consented person who is identifiable in a frame means the footage is de-identified or discarded, so settle bystander handling in the spec. On safety and qualification, the site states no contributor safety protocol, no guidance on protective equipment, and no safety or trade certification for any woodworking row, and this page describes none. The site lists capture at your own facility as not offered. None of this is legal advice, and the executed agreement controls.

Scale and speed

How do custom, fast and on-demand collection apply to wood cutting?

This is custom collection: custom collection at scale through a network of thousands of capture providers, for a consistent, recurring data flow to labs. You name the task, the setting and the variation your pipeline is missing, and capture is run to a written skill spec rather than drawn from a fixed catalog dataset.

From signed spec to first validated batch, the 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. The site does not say whether the wood cutting settings count as deep or thin coverage, so ask which applies when you request the quote. After the first batch, on-demand top-ups run against the same accept criteria. Collections are billed per validated hour, and the site publishes no price list, minimum order or batch size for this task family, so those are set in your quote.

Delivery

Which formats and license come with the data?

Delivery is in the format your stack already reads: RLDS, LeRobot, WebDataset, HDF5, zarr or Rerun .rrd, or raw streams with metadata, and the site also offers MCAP. Readable converter source ships alongside the data so the schema can be retargeted to your own loader. Topic names, message schemas and rates for any one container are not stated on the site and are agreed in the skill spec.

The license is perpetual, irrevocable and worldwide, with the data buyer-owned and non-exclusive by default. Training, evaluating, fine-tuning and deploying models, including commercial ones, is permitted, and exclusivity is available as a separate term. The executed agreement controls. The tradeoff to plan for is the embodiment gap between a human holding a workpiece at a saw and a robot platform doing the same, which retargeting or pairing with robot data has to close.

Ordering

How do you order an egocentric table saw and wood cutting dataset?

Describe the tasks and where each starts and ends: the measuring and marking, the table saw cut, the jigsaw shaping and the timber dimensioning steps, the start state of each workpiece, what counts as a reject, and the variation your pipeline is missing. Then name the streams you want, the delivery format, and whether calibration ships inside the file or beside it.

Most programs begin with a Spec Pilot, a small, fast first batch that proves the spec and accept criteria on real footage before you scale. The last table below shows what each spec element decides. For the full list of rows rated Hard, see the hard-task hub linked below.

At a glance

What you get.

Catalog rows
"Cut and shape wood panels using a table saw and jigsaw" and "Wood Cutting" in the Carpentry & Furniture Workshop environment of the Manufacturing & Workshops sector (both Hard)
Catalog descriptions
"Measure and mark panels, cut them on the table saw and shape curves with a jigsaw." and "Dimensioning and cutting timber with commercial saws and cutting equipment."
Status
Active collection on both rows
Catalog capture setup
Phone (head-mounted) / multiview rig on both rows
Streams in the wider stack
Ten: 4K head camera, two wrist cameras, depth, IMU, 3D hand and body pose, instance segmentation, action labels; aligned under 2 ms
Hand pose
21 joints per hand with a per-joint visibility flag
Labels
Action segments with start_ns, end_ns, verb and noun; contact events; instance masks at 848×480
Delivery
RLDS, LeRobot, WebDataset, HDF5, zarr, Rerun .rrd or raw streams with metadata; MCAP offered
License
Perpetual, irrevocable, buyer-owned, non-exclusive by default

Every collection runs through the same seven-stage end-to-end custom collection pipeline. End-to-end custom data collection is a managed service that takes an AI data need from problem to owned dataset in a single accountable pipeline.

Catalog rows

Which catalog rows cover table saw and wood cutting work?

The rows relied on by this page, with the catalog description, environment, skill, capture setup and status as listed.
Task (as listed)Description (as listed)EnvironmentSkillCapture setupStatus
Cut and shape wood panels using a table saw and jigsawMeasure and mark panels, cut them on the table saw and shape curves with a jigsaw.Carpentry & Furniture WorkshopHardPhone (head-mounted) / multiview rigActive collection
Wood CuttingDimensioning and cutting timber with commercial saws and cutting equipment.Carpentry & Furniture WorkshopHardPhone (head-mounted) / multiview rigActive collection

Compared pages

How does this page differ from the stereo and broader woodworking pages?

Each page, its main topic and how it treats the two workshop cutting rows. “Not covered” means the page leaves it out.
PageMain topicCapture framingTreatment of the two cutting rows
This page: egocentric table saw and wood cutting dataThe two workshop cutting rowsFirst-person, head-mounted phone or multiview rig as the catalog lists itStreams, labels, Hard rating, limits and ordering for these rows only
Stereo table saw and wood cutting dataThe same two rowsStereo pair, agreed in the skill spec; not stated on the rowsStereo and MCAP treatment, stream map and failure modes
Egocentric woodworking and carpentry dataThe eleven rows of the Carpentry & Furniture Workshop environmentFirst-person across all eleven rowsBoth rows among eleven; sanding, joining, hardware and finishing also covered
Hard task data collectionEvery catalog row rated Hard, grouped by environmentAs each row lists itBoth rows listed in the hub table with their status and capture setup

Stated vs not stated

What does the site state about egocentric capture for wood cutting?

Left to right: the item, what the site states, and what that means for a wood cutting order. “Not stated” means the site does not say.
ItemWhat the site statesFor a wood cutting order
Catalog capture setupPhone (head-mounted) / multiview rig on both rowsStated; which tier a program uses is set in the spec
Wider stackTen streams: 4K head camera, two wrist cameras, depth, IMU, 3D hand and body pose, instance segmentation, action labels; aligned under 2 msWhether a given order uses every stream: set in the spec
Hard-environments stack4K/60 head camera, 120° lens; two 1080p/60 wrist cameras; depth 848×480 at 30 fps; 200 Hz 6-DoF IMUWhether a wood cutting order uses it: Not stated
Hand pose21 joints per hand with a per-joint visibility flagStated
Calibration and syncTwice per session under a 0.28 px RMS gate; hardware-triggered onto one PTP clock domainStated
DepthPer-frame confidence maps; degrades on wet and reflective surfacesBehavior around sawdust or a moving tool: Not stated
Row statusBoth rows are Active collectionNo volume, hours or delivered footage is claimed on this page
Wood Cutting scopeCommercial saws and cutting equipment in a joinery and furniture workshopNo sawmill, industrial production line or mill capability is stated
Force or tactile streamsNot among the listed streams of standard captureNot claimed on this page
Wood species, panel dimensions, cut-line deviation or toleranceNo label statedAgreed in the skill spec, or not claimed on this page
Safety, protective equipment, saw models, cycle times, throughput or resultsNot stated; the rows say table saw, jigsaw, cutting and shaping wood panels, and cutting timber onlyAgreed in the skill spec, or not claimed on this page

Streams and labels

Which streams and labels can an episode carry?

Left to right: the stream or label, what the site states about it, and what it records for cutting work. “Not stated” means the site does not say.
Stream or labelWhat the site statesWhat it records for cutting work
Head camera4K head camera, first-person RGBThe panel, the line and the tool from the contributor’s point of view
Wrist camerasTwo wrist camerasA closer view of the hands than the head camera alone
Depth with confidence mapsPer-frame confidence maps; instance masks stored at 848×480Depth where confident; low-confidence regions masked
IMU6-DoF at 200 Hz from head- and wrist-mounted sensorsHead and wrist motion
3D hand posehand_joints and hand_visible; 21 joints per hand with a visibility flagBoth hands, with hidden fingers flagged
Contact eventscontact_events: the moment the hand meets an objectWhether a panel, timber or tool counts as the object: Not stated
Action segmentsstart_ns, end_ns, verb and nounOne labeled span per step as the spec defines it
Task instructionOne per episodeThe named task for the episode

Delivery and licensing

What is delivered, in which formats, and under what license?

Delivery and licensing terms as stated on this site. “Set per program” means the site states no fixed value.
ItemAs stated on this siteNote
FormatsRLDS, LeRobot, WebDataset, HDF5, zarr, Rerun .rrd, or raw streams with metadata; MCAP also offeredChosen in the skill spec; converter source ships with the data
LicensePerpetual, irrevocable, worldwide, buyer-owned, non-exclusive by defaultExclusivity is available as a separate term; the executed agreement controls
BillingPer validated hourRejected items are recaptured at our cost
PrivacyFaces, plates and screens blurred irreversibly; footage that cannot be de-identified is discardedAnonymization record delivered with the data
Timelinea median of 48 hours to first validated batch where coverage is deep; three to four weeks where thinWhich applies to wood cutting is Not stated; quoted against your spec
Price, minimum order, batch sizeNot published for this task familySet in your quote

Skill spec

What does the skill spec fix for a wood cutting order?

Spec element and where a buyer can see it in the delivery. “Set per program” means the site states no fixed value.
Spec elementWhat it decidesWhere it shows up
Tasks and orderWhat one step is: measure and mark, cut on the table saw, shape with a jigsaw or dimension timberEpisode boundaries and action segments; set per program
WorkpiecesWhich panels or timber pieces and how each beginsSet per program; both rows are Hard and Active collection
Tools and equipmentWhich saws and cutting equipment are namedSet per program; the site states no saw model
Episode boundaryOne cut, one panel or a full pieceSet per program; the site does not state an episode length for these rows
Capture tier and streamsHead-mounted phone or multiview rig, wrist views, hand pose, IMU and any othersThe streams you can read; confirm on the Spec Pilot
Accept criteriaWhat passes QA, including what counts as a rejectReject log per batch
Delivery formatHow the data is packagedOne of the listed formats, or several
LicenseWho owns the datasetPerpetual, irrevocable, buyer-owned, non-exclusive by default

Example footage

What does egocentric stereo capture look like?

One real clip already published on this site, shown as a format and rig example: a head-mounted capture with hands in frame. It does not show table saw, jigsaw or wood cutting work.

egocentric_stereo_sample.mp4 · head-mounted stereo · hands in frame · 1920 × 1080 web preview · rig example, not a wood cutting clip

No table saw, jigsaw or wood cutting clip is published on the site, so this is labelled as a format and rig example. Request a sample to see delivered files. See sample data outputs, request an egocentric data sample.

FAQ

Egocentric table saw data, answered.

What is egocentric table saw and wood cutting data?
First-person video of a person measuring and marking wood panels, cutting them on a table saw, shaping curves with a jigsaw, and dimensioning and cutting timber, recorded from their own point of view with hand pose and synchronized sensor streams alongside.
Which catalog rows cover it, and what is their status?
Two rows in the Carpentry & Furniture Workshop environment: "Cut and shape wood panels using a table saw and jigsaw" and "Wood Cutting". Both are rated Hard, list a head-mounted phone or multiview rig, and are Active collection. This page claims no volume or hours.
What does Hard mean here?
Only the skill level the catalog lists in its Skill column, one of Easy, Medium or Hard. The site states no failure rate, tolerance or worker skill claim for these rows.
How is this different from the stereo table saw page?
Both cover the same two rows. This page treats them as first-person capture, with the streams, labels and limits the site states. The stereo page treats them as a stereo pair and MCAP question; a stereo pair is agreed in the skill spec and is not stated on the rows.
Is there an off-the-shelf wood cutting dataset, and does this mean a sawmill?
The site lists no off-the-shelf table saw or wood cutting dataset, so this is custom collection, at scale through a network of thousands of capture providers, for a consistent, recurring data flow to labs. The site states no sawmill, industrial production line or mill capability.
How fast can a custom wood cutting dataset be collected?
From signed spec to first validated batch, a median of 48 hours where coverage is deep and three to four weeks where it is thin. The site does not say which applies to wood cutting settings, so ask when you request the quote.
What labels come with the data, and what is not labeled?
Action segments with start_ns, end_ns, verb and noun, 3D hand pose with a per-joint visibility flag, contact events and instance masks. There is no stated label for wood species, panel dimensions or cut tolerance, and no force or tactile stream.
Which formats are delivered, and who owns the data?
RLDS, LeRobot, WebDataset, HDF5, zarr or Rerun .rrd, or raw streams with metadata, and MCAP is also offered. The license is perpetual, irrevocable and worldwide, buyer-owned and non-exclusive by default.

Sources

Where are these capture and format claims documented?

Request a custom egocentric table saw and wood cutting dataset.

Tell us the tasks and where each starts and ends, the panels or timber and how they begin, the setting, and the streams and format you want. We will quote reach, timeline and price against a written spec.