Sourcing mocap human demonstrations

Quick facts

AMASS

Archive of Motion Capture as Surface Shapes — pools 15+ mocap datasets in a unified SMPL skeleton (Mahmood et al., ICCV 2019).

Aria — Nymeria

Project Aria-driven multimodal motion dataset combining body pose, gaze, and language; release via the Aria research-partner program.

HOI4D

4D hand-object interaction dataset — fine-grained hand pose plus object contact annotations from egocentric capture.

Why custom capture

Humanoid retargeting needs the buyer's robot's degrees of freedom and joint limits; off-the-shelf mocap rarely encodes target-embodiment constraints.

Spec checklist

Skeleton schema (SMPL/SMPL-X/custom), marker set, joint-rate (≥120 Hz typical), object-contact labels, performer consent for commercial training, retargeting reference video.

Comparison

Dataset requirements

Buyers should specify body, hand, and object motion capture with video reference, accepted scenes in studio or controlled workspace task capture, a minimum of 10 hours of useful captured volume, consent rules, and the exact metadata package: skeleton schema, marker set, task segment, object contact, and performer consent ^[1]. The Datasheets framework spells out which dataset-documentation questions matter before any commercial training program begins.

"The machine learning community currently has no standardized process for documenting datasets, which can lead to severe consequences in high-stakes domains."

Best-fit buyers

The strongest fit is humanoid and dexterous manipulation teams needing motion priors ^[3]. It can also work as a smaller eval set — typically 100 to 500 episodes — before a larger net-new capture program.

Mocap human demonstrations sample package

A credible mocap human demonstrations supplier should provide a sample package that includes raw files, a manifest, capture context, and these critical metadata fields: skeleton schema, marker set, task segment, object contact, and performer consent ^[4]. Across at least 10 representative episodes, the buyer should be able to inspect whether body, hand, and object motion capture with video reference actually appears in studio or controlled workspace task capture, not just trust a verbal description of the inventory.

Mocap human demonstrations licensing check

The licensing review for mocap human demonstration dataset should confirm whether the data is off-the-shelf or net-new, whether it can be used for commercial model training, whether contributors or sites consented, and whether the supplier can reproduce the same rights package for the full delivery ^[5]. A well-scoped buyer typically reviews 3 to 5 supplier deliveries before approving scale; without those checks, an apparently useful dataset can become a legal or procurement blocker.

Related pages

Use these to move from category-level context into specific task, dataset, format, and comparison detail.

External references and source context

1. Datasheets for Datasets

   Supports the dataset-requirements framework dimensions: dataset motivation, composition, collection process, recommended uses, and license review.

   arXiv ↩
2. Datasheets for Datasets

   Datasheets for Datasets defines the consent, provenance, and intended-use questions buyers must ask before commercial training; quoted verbatim in the dataset-requirements section.

   arXiv ↩
3. Open X-Embodiment: Robotic Learning Datasets and RT-X Models

   Open X-Embodiment establishes the cross-embodiment robotics pretraining baseline — a useful reference for buyer fit when mapping a deployment-specific dataset onto a generalist policy.

   arXiv ↩
4. Data Cards: Purposeful and Transparent Dataset Documentation for Responsible AI

   Data Cards capture dataset origins, development, intent, and ethical considerations buyers can attach to each delivered batch for procurement audit.

   arXiv ↩
5. encord

   Commercial vendors deliver licensed dataset collection programs with explicit contributor consent, rights, and per-batch documentation buyers can audit before scale.

   encord.com ↩

FAQ