Creating Art with Generative Adversarial Networks
Implementations of DC-GANs and WGANs in order to generate Art. Art images from Wikiart. MNIST and CIFAR-10 are also avalaible as toy datasets. The networks are trained to not only learn the dataset distribution, but also be able to chose the art style generated.
The network is asked to generate 2 art styles at the same time.
Run python main_mnist.py for DC-GAN or python main_mnist_wgan.py for wgan.
Run python main_cifar.py for DC-GAN or python main_cifar_wgan.py for wgan.
Download wikipaintings at www.lamsade.dauphine.fr/~bnegrevergne/webpage/software/rasta/wikipaintings_full.tgz.
Update the data path in main_art_wgan.py
Run the command python main_art_wgan.py