Dataset · BitRobot
100 hours of first-person LiDAR video for robots
BitRobot/RoboCap-LiDAR-100
Head-mounted camera and LiDAR scanner recordings of people working in kitchens and auto repair shops in Southeast Asia.
What's inside
Inside are 100 hours of first-person video (from 1,000 to 10,000 records) released in two 50-hour parts with English descriptions. Each episode contains a head-mounted camera stream, a LiDAR depth map (320x240), and a colored point cloud. Data is collected in real-world conditions: restaurant kitchens, an auto repair shop, and a detailing center in Southeast Asia.
What you can do with it
- Training depth estimation and 3D reconstruction models from video and LiDAR
- Training robots on complex domestic and repair manipulations demonstrated by humans
- First-person action classification in work scenarios
- Tasks
- Robotics, Depth estimation, Video classification
- Languages
- English
- Size
- 1K–10K rows
- Downloads this month
- 2.8K
- Likes
- 4
- Access
- on request on Hugging Face
How to load it
from datasets import load_dataset
ds = load_dataset("BitRobot/RoboCap-LiDAR-100")