Skip to content

Dataset · chcorbi

Panoramic 360 Stereo Pairs for Depth Estimation

chcorbi/helvipad

39,553 real-world panoramic stereo images with depth maps and LiDAR disparity.

What's inside

The HELVIPAD real-world stereo dataset includes 39,553 pairs of equirectangular 360-degree images captured by top and bottom cameras. The images are accompanied by pixel-level depth and disparity labels obtained from LiDAR point clouds. Filming took place in various indoor and outdoor locations under different lighting conditions, including nighttime.

What you can do with it

  • Train depth estimation models for panoramic cameras
  • Develop spatial perception systems for robotics and VR
Tasks
Depth estimation
Size
10K–100K rows
Downloads this month
52.4K
Likes
12

How to load it

from datasets import load_dataset

ds = load_dataset("chcorbi/helvipad")