Skip to content

Dataset · genrobot2025

Egocentric video and SLAM data for robotics

genrobot2025/Gen-EgoData

Demo sample of 500 first-person videos (4.23 hours) with motion trajectories for training home robots.

What's inside

An introductory snippet of the Gen EgoData dataset containing 500 first-person videos with a total duration of 4.23 hours with English and Chinese descriptions. Videos cover 10 household scenarios (50 examples per task) and are supplemented with camera positioning data (ego-SLAM) generated by the authors' algorithm.

What you can do with it

  • Training robotics and reinforcement learning models to perform first-person household actions
  • Testing visual odometry, SLAM, and trajectory prediction algorithms
Tasks
Robotics, Reinforcement learning
Languages
English, Chinese
Size
100B–1T rows
Downloads this month
3.6K
Likes
38

How to load it

from datasets import load_dataset

ds = load_dataset("genrobot2025/Gen-EgoData")