Ego4D is a large-scale first-person video dataset that contains more than 3,025 hours of video recorded from 73 different locations in 9 countries, with a total of 855 people.

Ego4D is the largest first-person video dataset of everyday activities. Some footage also includes audio, data about where the participant’s gaze is focused, and multiple perspectives of the same scene.

This dataset also introduces new benchmark challenges:

• Episodic Memory: Where is my X?
• Hand-object interaction: How do objects change during interaction?
• Audiovisual diary: Who said what and when?
• Social interaction: Who is interacting with whom?
• Prediction: What will happen next?