Egocentric HVAC Service Video Data

Egocentric HVAC service data is real-world video of a technician doing an HVAC service call while wearing the capture rig: inspecting the unit, replacing filters or parts, tuning the system and testing it. The footage shows the hands, the opened unit and the part being worked on from the head-mounted viewpoint. Each episode follows the Firsthand episode schema, with ten streams synced under 2 ms. Which streams a head-mounted phone populates, and which the rig adds, is confirmed in the spec.

Teams search for it as HVAC technician training data, an HVAC service video dataset, egocentric trade service call data and custom field task data collection. They all describe demonstrations of a service call, collected to a written spec. This page covers only what the Firsthand site states and marks everything else as Not stated, including any claim about refrigerant handling, certification, safety or code compliance.

The site states five HVAC-related rows in its commercial task catalog. Four of them name HVAC directly: "HVAC service call", "Filter Swap & System Tune-Up", "HVAC Equipment Installation" and "Replace and tune HVAC filters using screwdrivers and gauge sets". A fifth, "Thermostat & Control Wiring", names heating and cooling calls without using the word HVAC. Two rows are rated Medium and three Hard.

The rows split across two environments. The service call sits in On-Site Home Repair, and the other four sit in Trade Service Calls. Anything outside the stated wording is Not stated, including the system type, the unit size and the fault. A policy that needs another HVAC task writes it into the spec and we scope it, because a catalog row is a starting vocabulary, not a fixed menu.

The row reads "Inspect the HVAC unit, replace filters or parts, check refrigerant and test the system". It is rated Hard, its capture mode is listed as phone (head-mounted) or multiview rig, and its status is Available on request. It belongs to On-Site Home Repair, an environment described as "Professional plumber, electrician, HVAC technician, or handyman working in a client’s home".

The site defines that status plainly: the task is scoped and recruited to spec when a buyer asks for it. So recording for this row is not already running, and a first batch is quoted against your spec. The wording runs in order: inspect, replace, check, test. It states no unit type, no part, no duration, and no city, region or client.

Two rows in Trade Service Calls cover filters. "Filter Swap & System Tune-Up" reads "Replace HVAC filters, clean coils, check pressures and tune the system". The longer-titled filter row reads "Open the unit, swap the filters, connect gauges and adjust settings to spec". Both are rated Medium, below the Hard service call, and both are Ready to deploy.

The site defines Ready to deploy as contributors and site access set up, with capture starting once a spec is signed. The two rows overlap on swapping filters and differ on the rest: one adds coil cleaning and a pressure check, the other adds opening the unit and adjusting settings. Neither states a unit type, a target setting or a result, and this page does not infer one.

Yes, in the same Trade Service Calls environment, both Hard and Ready to deploy. "HVAC Equipment Installation" reads "Position and mount HVAC units, connect ducting, refrigerant lines and wiring, then commission". "Thermostat & Control Wiring" reads "Mount the thermostat, connect control wires to terminals and test the heating and cooling calls".

Installation is a different job from a service call, so this page treats it as an adjacent row the site states, not as the same task. It names refrigerant lines only as one of the things connected, and the page adds no procedure, certification or licensing claim. Both rows share the capture modes and the episode schema with the service rows.

It depends on the row. The HVAC service call is in On-Site Home Repair, a technician working in a client’s home. The other four rows are in Trade Service Calls, described as "Plumber, electrician or HVAC crew on service calls at client sites". That environment says client sites, and it does not say residence, so this page does not claim in-home capture for those four rows.

Neither description is a staged set; both are professionals doing real work for a client. The site states that egocentric capture happens in real kitchens, workplaces and streets, names no city, region, client or address, and states no number of sites or episodes. Environment, access and schedule are agreed in the written spec.

In the phone mode, a head-mounted phone records first-person video of real work on site. In the rig mode, a multiview rig adds synchronized third-person angles when the spec asks for them. Every HVAC row lists both, and the catalog does not say which applies to which row. Camera count, resolution, frame rate, mounting position and tool list are Not stated beyond the wording of each row.

Each episode follows the Firsthand episode schema: ten streams synced under 2 ms, with RGB, depth, 3D hand pose with per-joint visibility, contact and grasp events, and frame-accurate action segments. The site also states a 200 Hz 6-DoF IMU among its sensors. For HVAC, action segments are where a spec marks each step: the inspection, the part or filter swap, the adjustment and the test. Contact events mark when a hand meets a part, a panel or a control.

Not stated: whether depth and hand pose are populated in the phone mode, and whether a repair or test result is labeled as a success or quality field. Task duration, the number of episodes or sites, and robot results are Not stated too.

Three rows mention refrigerant: the service call says "check refrigerant", and the installation row says "refrigerant lines". The filter row mentions "check pressures". This page quotes those words and goes no further. It gives no handling procedure, no certification or licensing statement, and no regulatory, safety or code-compliance guidance, and the site gives none for these rows.

Those matters are set by the trade’s own standards and the client’s own site rules, not by a data page. What a collection spec can do is name the steps a buyer wants captured and the accept criteria each batch is checked against.

The site gives no technical analysis of HVAC, so this page makes no claim about robot results. What the row wording shows is an ordered arc rather than a single grasp: inspect, replace, tune and test. A service call also changes the object under the hands, because the unit is opened and closed and a part is swapped, which is a different data shape from repeating one motion at a bench.

The site rates two of the HVAC-named rows Medium and the rest Hard, which is the only statement on difficulty. It states no success or quality label, so this page makes no claim that any row demonstrates a service call done correctly, only that each row names a step sequence a spec can capture.

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

Every batch is validated against signed accept criteria before it ships, with a reject log, and delivery is rolling, per validated batch. Rejected items are recaptured at our cost, and billing is per validated hour.

Every episode links to a signed contributor release through a documented consent chain, and the consent record is part of the delivery. The release also covers bystanders: contributors confirm they have not knowingly recorded anyone who has not signed it, and where an unidentified person appears, the footage is de-identified or discarded.

A vision model runs on every frame at ingest, flagging faces, heads, licence plates and any legible screen or document. Those regions are irreversibly blurred, and a human reviewer confirms the result before footage ships. If the reviewer is unsure a region is fully covered, the episode is held or discarded, never downgraded and shipped. Access to a client’s home or site is a condition the spec sets, so raise any resident or site-access need before the spec is signed.

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. The four Trade Service Calls rows are Ready to deploy, so contributors and site access are set up. The service call row is Available on request, so it is recruited to your spec first.

The site also states program milestones. A pilot reaches its first delivery in two to three weeks, and a standing program reaches full rate in four to six weeks. Those assume capture coverage is already in place. A batch size and a later-batch turnaround are Not stated, so both are quoted against your spec.

You describe the service calls, the steps and the parts 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. This is the custom egocentric data collection route that every Firsthand task family follows.

A Spec Pilot, a small and fast first batch, can prove the spec and accept criteria before you scale. Every collection ships under a perpetual, buyer-owned commercial license, with a reject log and signed accept criteria on every batch.