This is codes for Unsupervised Diverse Colorization via Generative Adversarial Networks
tensorflow: 1.0+
numpy
scipy
Pillow
cv2
Download LSUN/bedroom_train data.
Maximum center crop the data and reshape into 64*64,
put reshaped data into data/lsun_64/bedroom_train/
run train_lsun_wgan.sh for shortcut
or adjust settings by:
python main_wgan.py --is_train=True --some_paramter=some_value
run test_lsun_wgan.sh for shortcut
or adjust settings by:
python main_wgan.py --is_train=False --some_paramter=some_value
Some demo results in demo/gan/