Nymeria

Meta's large-scale first-person dataset of everyday human motion, recorded with Aria glasses and a motion-capture suit.

Nymeria is a first-person human-motion dataset released by Meta's Project Aria team, published at ECCV 2024. 264 participants performed everyday activities at 50 real-world locations, totaling about 300 hours: Aria smart glasses on the head recorded first-person video, eye gaze, and IMU data; a miniAria wristband was worn on the wrist; an XSens inertial motion-capture suit provided full-body ground-truth motion; observers filmed a third-person view as well, and everything is paired with hierarchical language descriptions. It fills a data gap around reconstructing what a person's entire body is doing from head-worn devices alone, and can be used for full-body pose estimation, motion generation, and action recognition. It's released under CC BY-NC 4.0 (non-commercial); an expanded version, NymeriaPlus, followed later.

ExampleFirst-person video and IMU signals recorded with Aria glasses are used to train a model to estimate the wearer's full-body pose, with the synchronized XSens motion-capture data serving as ground truth to measure the error.

Also called

Nymeria Dataset, NymeriaPlus

Related

Egocentric Video · Project Aria Glasses · Inertial Motion Capture · Ego-Exo4D · Human Video Data · Motion Capture

Sources

Nymeria: A Massive Collection of Multimodal Egocentric Daily Motion in the Wild (arXiv 2406.09905)
Project Aria - Nymeria Dataset
facebookresearch/nymeria_dataset (GitHub)

As of

2026-06