Photometric Stereo via Discrete Hypothesis-and-Test Search

CVPR 2020

Kenji Enomoto¹ Michael Waechter¹ Kiriakos N. Kutulakos² Yasuyuki Matsushita^1
1Osaka University ²University of Toronto

Paper   Supplementary

---

This is the official implementation for our CVPR paper 'Photometric Stereo via Discrete Hypothesis-and-Test Search'.

How to use

# Docker
docker build -t hat-search-ps:latest -f Dockerfile .
docker run --rm -itd --name=hat -v /path/to/any/dir:/mnt -v /path/to/hat-search-ps-cvpr2020:/hat-search-ps-cvpr2020 hat-search-ps:latest
docker exec -it hat bash
cd /hat-search-ps-cvpr2020

# Download MERL database & DiLiGenT dataset
mkdir /mnt/merl_brdf_database
sh script/download_merl_brdf_database.sh /mnt/merl_brdf_database
sh script/download_diligent_dataset.sh /mnt

# Create matrix D
python3 create_matrix_D_merl.py --N_file supp_info/N_05_interval.txt --L_file supp_info/L_diligent1.txt --merl_dir /mnt/merl_brdf_database --material_name_file supp_info/merl_name.txt --out_dir /mnt/matrix_D/diligent_1

# Compute matrix Z
python3 compute_Z_from_D.py --D_dir /mnt/matrix_D/diligent_1 --rank 3 --out_dir /mnt/matrix_Z/diligent_1/rank_3 --n_jobs -1

# Evaluate
python3 run.py --root_dir /mnt/DiLiGenT/pmsData --obj_file supp_info/diligent_name_1.txt --data_name diligent --Z_dir /mnt/matrix_Z/diligent_1/rank_3 --N_file supp_info/N_05_interval.txt --out_dir /mnt/result/diligent --n_jobs -1

Citation

@inproceedings{enomoto2020photometric,
    title = {Photometric Stereo via Discrete Hypothesis-and-Test Search},
    author = {Kenji Enomoto and Michael Waechter and Kiriakos N. Kutulakos and Yasuyuki Matsushita},
    booktitle = {Computer Vision and Pattern Recognition (CVPR)},
    year = {2020}
}