Dataset / arXiv / Oct 13, 2021
Ego4D: Around the World in 3,000 Hours of Egocentric Video
Ego4D introduced a large-scale first-person video dataset and benchmark suite that helped establish the breadth of egocentric visual understanding tasks.
Ego4D brings together a large and geographically diverse collection of first-person video for studying activities, interactions, memory, and social understanding.
The dataset remains useful product context because it demonstrates how capture consistency, wearer comfort, and broad scene coverage shape long-form egocentric research.
The source paper contains the complete dataset design, benchmark tasks, consortium information, and research findings.
egocentric visiondatasetactivity understandingfirst-person video
Open original source