Skip to content

Dataset · eidon-ai

Egocentric Household Video with IMU Hand Tracking

eidon-ai/tracker-pov

1,274 hours of first-person video with synchronized 7-point IMU hand tracking during household chores.

What's inside

The dataset contains 1,274 hours of egocentric video recorded via a head-mounted camera during routine housework: laundry, cleaning, dishwashing, and cooking. Each video comes with orientation data from seven IMU sensors at 24 Hz for both hands, forearms, shoulders, and chest. This is the final full release; participants wore a seven-sensor IMU system and a camera in their own homes.

What you can do with it

  • Train an action classification model on egocentric video
  • Verify video- and IMU-based hand pose estimation
  • Fine-tune a robotic policy on household actions
Tasks
Robotics, Video classification
Size
10K–100K rows
Downloads this month
69.9K
Likes
81

How to load it

from datasets import load_dataset

ds = load_dataset("eidon-ai/tracker-pov")