Off-the-shelf · Household

Off-the-Shelf Household Egocentric Video Data

Last reviewed:

Short answer

Off-the-shelf household egocentric video is first-person footage of real chores, in stock and ready to license. The library lists 20,252+ hours across two datasets: a 20,000-hour smartphone set of cleaning, organizing, laundry and groceries, and a 252-hour annotated stereo set from homes and commercial spaces. Pricing is a custom quote.

At a glance

What is in stock.

Household data is the deepest pool in the library, and the fastest way to put real home manipulation in front of a model without waiting for a new collection.

Datasets
05 Smartphone, household · 01 Stereo, annotated
Listed volume
20,252+ hours
Tasks
Cleaning, organizing, laundry, groceries (05); household and commercial (01)
Devices
Smartphone (05); head-mounted stereo rig with dense action labels (01)
IMU
No (05); yes (01)
Regions
Indonesia 95%, Philippines 2.5%, Kenya 2.5% (05); India (01)
Licence
Non-exclusive, with resale rights
Samples
Sample folder and full metadata sheet per dataset, before you buy
Pricing
Custom quote from a rep, based on volume, mix and licence terms

Inventory summary. Exact delivery contents are confirmed per order.

Datasets in stock

Which datasets are available?

Each dataset has a sample folder and a full metadata sheet you can review before ordering.

Household egocentric video: datasets with device, environment, IMU, region, volume and licence terms
#DatasetDeviceEnvironmentIMURegionVolumeTerms
05Smartphone, householdSmartphoneCleaning, organizing, laundry, groceriesNoIndonesia 95%, Philippines 2.5%, Kenya 2.5%20,000 hrsNon-exclusive, resale rights
01Stereo, annotated (calibrated stereo, dense action labels)Head-mounted stereo rigHousehold + commercialYesIndia252 hrsNon-exclusive, resale rights

Pricing is a custom quote from a rep, based on hours ordered, dataset mix and licence terms.

What is household egocentric video data?

It is video recorded from the point of view of the person doing the work: the camera sees what their eyes see, including their hands and the objects they handle. For home robots, that viewpoint is close to what a robot sees when it reaches for a dish, folds a shirt or puts groceries away.

Which household dataset should I start with?

For large-scale household pretraining, dataset 05 is the natural starting point: 20,000 hours of everyday chores recorded on smartphones, mostly in Indonesia with smaller shares from the Philippines and Kenya.

If you need depth, motion data or labels, dataset 01 is smaller but richer: 252 hours from a calibrated, head-mounted stereo rig with dense action labels and IMU, recorded in India in household and commercial settings.

When is off-the-shelf household data enough?

It fits when broad coverage of everyday chores matters more than a specific home, object set or task list: pretraining, representation learning, and early experiments that need volume now. If you need particular tasks, homes, countries or extra sensors, custom collection fills the gap on the same rigs.

FAQ

Questions buyers ask.

01

How many hours of household egocentric video are in stock?

The library lists 20,252+ hours across datasets 05 (20,000 hours) and 01 (252 hours).

02

Does the household data include IMU?

Dataset 01 includes IMU. Dataset 05 was recorded on smartphones and does not.

03

Which countries is it recorded in?

Dataset 05 is 95% Indonesia, 2.5% Philippines and 2.5% Kenya. Dataset 01 is recorded in India.

04

Can I resell or redistribute it?

Datasets 01 and 05 are non-exclusive and come with resale rights.

05

What if I need a different country or task?

Custom collection runs across LATAM, Africa, Asia, the USA and Europe on the same rigs, so any gap in the library can be filled to your spec.

Talk to a rep

Request samples and a custom quote.

Tell us the datasets and hours, or the tour industries and volume, you need. A rep sends samples, a quote and the order form. Need something the library doesn't hold? Ask for a custom dataset.

  • 01 · Share the datasets, industries or hours you need
  • 02 · Review samples and metadata sheets
  • 03 · Receive a custom quote and the order form

Prefer data captured to your spec? Request a custom dataset.