Team Ai
Datasetpublic

SeokjunChoi/display-inverse-rendering-dataset

Display Inverse Rendering Dataset πŸ“„ Paper (ArXiv) 🌐 Project Page πŸ’» GitHub Repository Introduction This dataset is created for display inverse rendering, including multi-light stereo images captured by polarization cameras, and GT geometry (pixel-aligned point cloud and surface normals) scanned by high-precision 3D scanner. Structure DIR-basic: The basic version of the dataset released with the paper. It includes stereo polarized RAW images… See the full description on the dataset page: https://huggingface.co/datasets/SeokjunChoi/display-inverse-rendering-dataset.

sourceHugging Facecc-by-4.0updated 1y agoView on Hugging Face
1likes199downloads
Dataset Card

Display Inverse Rendering Dataset

Introduction

This dataset is created for display inverse rendering, including multi-light stereo images captured by polarization cameras, and GT geometry (pixel-aligned point cloud and surface normals) scanned by high-precision 3D scanner.

Structure

  • β€”DIR-basic: The basic version of the dataset released with the paper. It includes stereo polarized RAW images, RGB images from a reference view, and ground-truth surface normals and point clouds. All images are captured under a multi-light configuration projected through 16Γ—9 superpixels on the display.
  β”œβ”€β”€ A
  β”‚  β”œβ”€β”€GT_geometry (for reference(main) view)
  β”‚  β”‚  β”œβ”€β”€'normal.npy',
  β”‚  β”‚  β”œβ”€β”€'normal.png',
  β”‚  β”‚  β”œβ”€β”€'point_cloud_gt.npy'
  β”‚  β”œβ”€β”€main
  β”‚  β”‚  β”œβ”€β”€diffuseNspecular
  β”‚  β”‚  β”‚  β”œβ”€β”€'000 - 143.png',
  β”‚  β”‚  β”‚  β”œβ”€β”€'black.png',
  β”‚  β”‚  β”‚  β”œβ”€β”€'white.png',
  β”‚  β”‚  β”œβ”€β”€RAW_polar
  β”‚  β”‚  β”‚  β”œβ”€β”€'000 - 143_[SHUTTER_TIME].png',
  β”‚  β”‚  β”‚  β”œβ”€β”€'black_[SHUTTER_TIME].png',
  β”‚  β”‚  β”‚  β”œβ”€β”€'white_[SHUTTER_TIME].png',
  β”‚  β”œβ”€β”€side
  β”‚  β”‚  β”œβ”€β”€diffuseNspecular
  β”‚  β”‚  β”‚  β”œβ”€β”€'000 - 143.png',
  β”‚  β”‚  β”‚  β”œβ”€β”€'black.png',
  β”‚  β”‚  β”‚  β”œβ”€β”€'white.png',
  β”‚  β”‚  β”œβ”€β”€RAW_polar
  β”‚  β”‚  β”‚  β”œβ”€β”€'000 - 143_[SHUTTER_TIME].png',
  β”‚  β”‚  β”‚  β”œβ”€β”€'black_[SHUTTER_TIME].png',
  β”‚  β”‚  β”‚  β”œβ”€β”€'white_[SHUTTER_TIME].png',
  β”‚  β”œβ”€β”€mask.png
  β”‚  β”œβ”€β”€point_cloud.npy (unprojected pixel w.r.t. depth & focal length)
  • β€”DIR-pms: This dataset follows the DiLiGeNT format and has the same composition as DIR-basic. It provides multi-light RGB images from the reference view along with related information and the ground-truth normal maps.
  • β€”
  β”œβ”€β”€ A [Suffix (default "PNG")]
  β”‚  β”œβ”€β”€'000 - 143.png',
  β”‚  β”œβ”€β”€'filenames.txt',
  β”‚  β”œβ”€β”€'light_directions.txt'
  β”‚  β”œβ”€β”€'light_intensities.txt',
  β”‚  β”œβ”€β”€'mask.png'
  β”‚  β”œβ”€β”€'Normal_gt.mat'

TODO

  • β€”[x] ~~Release training code.~~
  • β€”[x] ~~Release Display Inverse Rendering (DIR) dataset.~~
  • β€”[ ] Release EXPANDED version of DIR datset (HDR).
  • β€”[ ] Release EXPANDED version of DIR datset (multi-distance).
SeokjunChoi/display-inverse-rendering-dataset Β· Team Ai