Skip to content

Dataset · andaba

Multimodal 3D human pose benchmark via 4D radar, cameras, and LiDAR

andaba/RT-Pose

A research dataset and benchmark for 3D human pose estimation and localization using 4D radar and sensors.

What's inside

A dataset from an ECCV 2024 conference paper combining calibrated raw radar ADC data, 4D radar tensors, stereo RGB images, and LiDAR point clouds. The corpus size ranges from 100,000 to 1 million records. The data is intended for keypoint detection and 3D human pose tasks.

What you can do with it

  • Training multimodal 3D Human Pose Estimation models
  • Developing human tracking and localization systems for robotics and autonomous vehicles in challenging conditions
  • Comparing computer vision and radar perception algorithms on a unified benchmark
Tasks
Keypoint detection
Size
100K–1M rows
Downloads this month
228.7K
Likes
30

How to load it

from datasets import load_dataset

ds = load_dataset("andaba/RT-Pose")