Egocentric Landscaping and Groundskeeping Video Data

Egocentric landscaping and groundskeeping data is real-world, first-person video of work on outdoor grounds. It is recorded from the worker viewpoint, with depth, 3D hand pose, contact events and frame-accurate action segments aligned on one hardware clock. The tasks it covers are the ones the site states in its task catalog: landscaping and groundskeeping on commercial property, landscaping and groundskeeping on a farm, and pruning and trimming hedges.

This is outdoor maintenance work done with hand and power tools. The plant care and gardening page covers tending individual plants, the tool use page covers hand-tool manipulation across many settings, and the car washing and detailing page covers another kind of upkeep work.

The site states grounds work in its task catalog under the Agriculture & Grounds sector. A Landscaping & Groundskeeping environment lists its examples as commercial grounds, parks, corporate campuses and golf courses. Two further rows sit in the Farm & Field Operations environment: a farm landscaping and groundskeeping row, and a hedge pruning and trimming row.

Every row uses the same capture mode, a head-mounted phone or a multiview rig. All three are rated Medium. Two are listed as Active collection and one as Ready to deploy.

The catalog row reads "Mowing, trimming, planting and general grounds maintenance on commercial property." It is the broadest of the three.

The row is rated Medium and listed as Active collection. Equipment, site and lawn sizes, plant species, the number of sites and the order of the work are Not stated.

The catalog row, named Agriculture Landscaping & Groundskeeping, reads "Mow, trim, plant and clear grounds and paths on the farm with hand and power tools." It adds two details the commercial row does not give: clearing paths, and the use of hand and power tools.

The row is rated Medium and listed as Ready to deploy. Which hand and power tools are used, and what a cleared path looks like, are Not stated.

The catalog task reads "Prune and trim hedges using hedge trimmers and loppers", and its description reads "Cut hedge tops and sides with trimmers, remove thick stems with loppers and clear cuttings." It covers three linked steps: cutting the tops and sides, removing thick stems, and clearing what was cut.

The row is rated Medium and listed as Active collection. Hedge species, heights, lengths and the tools used are Not stated beyond the two tool types the row names, and so is any success or quality label for a trimmed hedge.

Indoor plant care and watering, greenhouse and nursery tending, and harvesting produce sit on the plant care and gardening page and in other Agriculture & Grounds environments. The farm environment also lists planting and pruning rows of its own. This page lists them as neighbours and does not count them as landscaping rows.

Raking leaves, leaf blowing, edging lawns, weeding and mulching planting beds are tasks people often expect here, but the site states none of them as a catalog row.

The catalog describes work done with "hand and power tools", "hedge trimmers and loppers" and general grounds maintenance across commercial property, parks and campuses. The site does not state terrain, weather, season, lighting or plant properties, so this page makes no claim about them. What the site does state is what egocentric video puts into the training distribution: hands entering the frame from below, occlusion during grasp and motion blur during reach.

The site also states why a hand pose stream matters here. RGB alone leaves a model to infer 3D hand structure it cannot see reliably through occlusion and motion blur. Firsthand delivers 3D hand pose as handJoints[2][21][3] in metric units in the head-camera frame, with a visibility flag per joint, aligned to depth and RGB. Two hands of 21 joints each are tracked per frame, and the flag records which of them were hidden behind a tool or foliage.

The catalog places the rows in a farm environment and a commercial grounds environment. The site names no park, campus, golf course, farm or grounds crew, and it does not state capture at a customer own site or facility. Environment, access and schedule are agreed in the spec, and the number of sites is Not stated.

Every episode links to a signed contributor release through a documented consent chain. Faces, plates and screens are detected and irreversibly blurred at ingest, a reviewer confirms the result before anything ships, and the anonymization record is delivered with the data. An undocumented bystander is also a stated reject criterion, and the site states no further privacy rule specific to outdoor sites.

A head-mounted phone records the first-person view of real work on site, and a multiview rig adds synchronized third-person angles when the spec asks for them. Both are the capture modes the catalog lists for every grounds row.

The site states a head camera at 3840 × 2160 and 60 fps for its instrumented rig, but no camera count, resolution or frame rate for the grounds rows, so the capture tier is agreed in the spec.

Each episode follows the Firsthand episode schema: ten streams synced under 2 ms, with head and wrist RGB, depth, 3D hand pose with per-joint visibility, contact and grasp events, and frame-accurate action segments. For grounds work, hand pose tracks how the fingers hold a tool, contact events mark when fingers or a tool meet a plant or surface, and action segments mark each step.

Object labels use the terms your spec sets. Force or tactile data, and success or quality labels for a mown area, a cleared path or a trimmed hedge, are Not stated. They are defined in the spec if your policy needs them.

The site offers a tactile glove tier as a pilot-ready option. The gloves track the hands, the glove signal is time-synced to a head-mounted stereo camera by hardware timestamps and an IMU, each glove is calibrated every session, and the pipeline checks signal dropout and sync drift.

The site does not state that the glove tier has been used on any grounds row, and it does not state whether gloves fit work with power tools or outdoors. Treat force and tactile data as Not stated for these rows unless a spec adds the glove tier and agrees the fit with you.

Episodes are machine-checked at ingest against the stated reject criteria, listed in the table below, and a human reviewer then checks the result.

It does not state how often a plant or tool leaves the view outdoors, and it states no blur or sharpness rule. A rejected item goes to the reject log and is recaptured at our cost.

Delivery formats are RLDS, LeRobot and WebDataset, plus HDF5, zarr and Rerun .rrd. The format table below sets each against what its official documentation states. Landscaping episodes use the same schema as every other Firsthand collection, so a loader that reads another Firsthand dataset reads this one without changes.

Every batch is validated against signed accept criteria before it ships, with a reject log, and delivery is rolling, per validated batch.

Where coverage is already deep, the first validated batch lands in a median of 48 hours from signed spec. Thin environments typically take three to four weeks, because vetting operators there is the bottleneck, and we quote that up front.

On-demand top-ups run against the same spec and accept criteria after each training run. The site lists the grounds rows as Active collection or Ready to deploy and states no capture timeline for them beyond a quote against your spec.

You describe the areas, tasks and sequence your policy needs. We turn that into a written spec covering tasks, environments, capture tier, annotation depth and accept criteria, then quote reach, timeline and price before any capture begins.

A Spec Pilot, a small and fast first batch, can prove the spec and accept criteria before you scale. Billing is per validated hour, and every collection ships under a perpetual, buyer-owned commercial license.