Skip to content

Latest commit

 

History

History
56 lines (48 loc) · 3.7 KB

nLMVS-Real.md

File metadata and controls

56 lines (48 loc) · 3.7 KB

nLMVS-Real Dataset (Preprocessed)

This dataset is provided under the Creative Commons Attribution 4.0 International License (CC BY 4.0). If you use this dataset please cite our paper.

@InProceedings{Yamashita_2023_WACV,
    author    = {Kohei Yamashita and Yuto Enyo and Shohei Nobuhara and Ko Nishino},
    title     = {nLMVS-Net: Deep Non-Lambertian Multi-View Stereo},
    booktitle = {Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
    month     = {Jan},
    year      = {2023}
}

This dataset is organized as follows.

.
├── nlmvs-real
│   └── ${illum name}_${mat name}
│       ├── illumination.exr          - Environment map (captured using mirrored ball, not to be used in our experiments)
│       └── ${shape name}
│         ├── illumination.exr        - Environment map (captured using RICOH THETA Z1)
│         ├── mesh_to_world.npy       - Transformation matrix from coordinate system of mesh files in ./mesh_files to world coordinate system
│         ├── mesh_aligned.ply        - Ground truth 3D mesh file aligned with captured images (please see below to prepare this file)
│         ├── view-??.exr             - Linear HDR image
│         ├── view-??.jpg             - Tone-mapped SDR image (not to be used in our experiments)
│         ├── view-??_m.png           - Object segmentation mask
│         ├── view-??_d.npy           - Ground truth depth map
│         ├── view-??_n.npy           - Ground truth normal map
│         └── views.txt               - Text file containing intrinsic and extrinsic camera parameters
├── mesh_files
│   └── ${shape name}_processed.ply - Ground truth 3D mesh file (not aligned with captured images)
├── create_aligned_meshes.py        - Script to create ground truth 3D mesh models aligned with captured images.
├── preprocess_bunny.py             - Script to pre-process the Stanford Bunny model
├── preprocess_planck.py            - Script to pre-process the Max-Planck Bust model
├── script_for_bunny.mlx            - Meshlab script for mesh preprocessing
├── script_for_planck_1.mlx         - Meshlab script for mesh preprocessing
├── script_for_planck_2.mlx         - Meshlab script for mesh preprocessing
└── README.md                       - Describing license, data organization, and acknowledgement

Creating ground truth 3D mesh models

  1. Download bunny.obj and maxplanck.ply from McGuire Computer Graphics Archive and Suggestive Contour Gallery, respectively. Save them to ./mesh_files.
  2. Run python preprocess_bunny.py and python preprocess_planck.ply to preprocess the mesh files.
  3. Run python create_aligned_meshes.py.

Acknowledgement

This work was in part supported by JSPS 20H05951, 21H04893, JST JPMJCR20G7, JPMJSP2110, and RIKEN GRP. We also thank Shinsaku Hiura for his help in 3D printing.

Use of existing assets

We used the following 3D mesh models to create this dataset.