efficient and robust white balance algorithm
Try to implement google's white balance paper: (The training code is based on google's ffcc code using Matlab, https://github.com/google/ffcc ). I modified the code to get a simplified version with one kernel only for my own project.
Barron, Jonathan T. "Convolutional color constancy." In Proceedings of the IEEE International Conference on Computer Vision, pp. 379-387. 2015.
Barron, Jonathan T., and Yun-Ta Tsai. "Fast Fourier Color Constancy." arXiv preprint arXiv:1611.07596 (2016).
-
Use tensorflow as a tool for optimization. (I tried, but it seems that tensorflow optimizer can not solve this optimization problem well.)
-
Implement an optimization solver based on google's ffcc (the open source code of the second paper). The original code has many redundant code, I tried to re-implement a much more clean and easy-use version here.
-
A fast CUDA based white balance algorithm
-
download the training data and pre-trained model and extract to the root dir: link: https://pan.baidu.com/s/1jKeQWKm passwd: h2v5 link: https://drive.google.com/file/d/18d12bQQv_ftPwFS3SZx0nFW6muApW_rB/view?usp=sharing
-
training code is in ./matlab_training
-
C++/CUDA code used to apply auto white balance on input image is in ./Cpp
-
pre-trained model is in ./data/model (you should download it from baiduyun)