Wugengxian / Image-color

Image Colorization of pix2pix with referenced graph

1. train.py and test.py

Use this to train the modified network.

python train.py --training_dir <the datasets> --preference

--preference means whether you need to train with reference image

Use this this to train the pix2pix without reference image

python train.py --training_dir <the datasets> --gan_g

2. Model directory

this directory is for you to obtain the pix2pix network framework.

3. Process_for_GAN directory

this directory is for you to train the model of origin net based on essay Colorful Image Colorization, skip connection model, GAN model and cycle GAN model. For more information, step into this directory for more details.