Learning to Separate Multiple Illuminants in a Single Image

This is the implementation described in the paper "Learning to Separate Multiple Illuminants in a Single Image, Zhuo Hui, Ayan Chakrabarti, Kalyan Sunkavalli, Aswin C. Sankaranarayanan, CVPR 2019" .

Website: https://huizhuo1987.github.io/learningIllum.html

The code skeleton is based on "https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix" and "https://github.com/lixx2938/CGIntrinsics". If you use our code for academic purposes, please consider citing:

@inproceedings{hui2019learning,
  	title={Learning to Separate Multiple Illuminants in a Single Image},
  	author={Hui, Zhuo and Chakrabarti, Ayan and Sunkavalli, Kalyan and Sankaranarayanan, Aswin C},
  	booktitle={Computer Vision and Pattern Recognition (CVPR 2019)},
  	year={2019}
}

Training dataset:

Download the dataset from Google drive: https://drive.google.com/file/d/1vcXDEKEQ_Id-ote0tfn4RDEF2fw6fzYs/view?usp=sharing

Test images:

Download the test image dataset from Google drive: https://drive.google.com/file/d/1MIPR3bvOVf-Bvcltu6-LZcTl7-sgl93U/view?usp=sharing

Pretrained model:

Download the pretrained the model: https://www.dropbox.com/s/cn1xylahysyqmnr/pretrained_models.zip?dl=0

Train the network

To train your network, run the following command

    python train.py --dataroot {path_to_training_data} --model threelayers --name {your_training_name} 
    --lrA 0.0001 --lrB 0.0001 --niter 100 --niter_decay 100 --display_id -1 --gpu_ids {your_gpu_ids}

Test image

To test the performance, run the following command

    python test.py --dataroot {path_to_test_data} --model threelayers --name {your_training_name} 
    --gpu_ids {your_gpu_ids}