IRIS offers public access to materials, data, and analysis tools for research on language learning, multilingualism, and language use.
Dataset Download
Ego-Exo4D is a multimodal video dataset and benchmark for studying skilled everyday activities from first and third person views.

About Dataset Download
Ego-Exo4D is a large-scale multimodal video dataset and benchmark focused on skilled human activities, captured simultaneously from first-person (ego) and third-person (exo) perspectives. It was collected across 13 cities by 740 participants, totaling 1286.3 hours of footage covering 9 skilled activity scenarios including cooking, basketball, bike repair, bouldering, music, soccer, health, dance, and rock climbing. The dataset includes three types of synchronized natural language annotations paired with video to support video-language research, multi-view capture using calibrated, time-synchronized Aria glasses and 4-5 stationary GoPros per scenario, and additional sensor modalities including microphones, IMUs, barometers, magnetometers, 6DoF localization data, eye gaze tracking, and point clouds. It supports research in multi-modal machine perception for daily life activity understanding, and hosts associated benchmark challenges in collaboration with eval.ai.
Key features
- Multimodal video dataset of skilled human activities from ego and exo views
- Three synchronized natural language annotation types paired with video
- Calibrated, time-synchronized multi-view capture with Aria glasses and GoPros
- Additional sensor modalities including IMUs, microphones, gaze, and point clouds
- Benchmark challenges for keystep recognition, proficiency estimation, and pose estimation
- Publicly accessible via CLI downloader after 2-day license approval process
- Large-scale dataset with over 1286 hours of synchronized ego-exo video across 9 skilled activity scenarios
- Includes three types of synchronized natural language annotations paired with video for video-language research
- Captures multiple additional sensor modalities alongside video to support multi-modal perception research
- Nearly all video is not de-identified, limiting use cases requiring anonymized participant data
- Access requires a 2-day license approval process, and prior Ego4D license signers must sign a new license
- Additional metadata, features, and models are not yet released, with planned availability before the end of the year
Dataset Download at a glance
- Website
- ego-exo4d-data.org
- Category
- Research & Data › Open Data & Datasets
- Type
- Website
- Pricing
- Not confirmed. Check the site
- Platforms
- Web
- Open source
- Not stated
- Language
- English
- Last checked
- Oct 6, 2026 · Online when checked
- Listed
- Oct 5, 2026
Dataset Download alternatives
Tools and sites you could use instead, ranked by how closely they match.
FRED is a freshwater research and environmental database with search and map views for study sites, samples, parameters, and species groups.
MovieLens 1M is a benchmark dataset containing one million movie ratings from 6,000 users.
The MUSE-Wide project page describes a survey of well-studied deep fields and links to its first data release.
CARS shares data from the first release of the Close AGN Reference Survey, a spatial survey of active galaxies.
The Wissenschaftliche Sammlungen portal helps users search university and college research collections, digital collections, and related activities.
Popular in Open Data & Datasets
Well-known listings from the same topic.
Dataset Download FAQ
How can I download the Ego-Exo4D dataset?
Do prior Ego4D license signers need to sign a new license for Ego-Exo4D?
Does the Ego-Exo4D dataset contain identifying information of individuals?
Tell us and an editor will check it.
Claim the listing and suggest updates yourself. An editor checks every claim first.











