Code of Neural Inverse Rendering for General Reflectance Photometric Stereo (ICML 2018)
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
chainer_code
data/DiLiGenT
README.md
overview.jpg

README.md

Neural Inverse Rendering for General Reflectance Photometric Stereo (ICML 2018)

Overview

We provide an implementation of the photometric stereo method presented in the following paper. The code can be used for research purpose only. If you find our code useful for your research, please cite our paper.

@inproceedings{Taniai18,
  author    = {Tatsunori Taniai and
               Takanori Maehara},
  title     = {{Neural Inverse Rendering for General Reflectance Photometric Stereo}},
  booktitle = {{Proceedings of the 35th International Conference on Machine Learning (ICML)}},
  pages     = {4864--4873},
  year      = {2018},
}

Links [Paper] [Project]

Running Environments

  • Python 3.5+
  • Chainer 3.4 (or maybe higher version is acceptable)
  • numpy
  • CUDA and cuDNN (if use the GPU mode "-g 0")

How to Run?

Move to chainer_code directory and enter the following command.

For CPU mode

python train.py -t 0

For GPU dode

python train.py -t 0 -g 0

Options

Options Type Settings
--gpu (-g) int GPU ID to use
--target (-t) int (0-9) Scene number from 0 to 9 (ball to reading)