Dataset · Hubdata
First-person video with manual work and IMU data
Hubdata/HubEgo-100K
Over 10,000 hours of first-person real-world workflows with motion sensors for training robots and multimodal models.
What's inside
The first release of HubEgo-100K contains 10,124 hours of unscripted first-person RGB video and inertial measurement unit (IMU) data recorded by 12,227 participants in real workplaces including construction sites, workshops, kitchens, salons, and homes. The data is provided in LeRobot and MCAP formats (10k to 100k records) with English annotations.
What you can do with it
- Training robots and VLA models for manipulation and complex manual skills
- Training video understanding and action recognition models (video-text-to-text, video-classification)
- Tasks
- Robotics, Video classification, Video text → text
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 335
- Likes
- 5
- Access
- on request on Hugging Face
How to load it
from datasets import load_dataset
ds = load_dataset("Hubdata/HubEgo-100K")