In this repo you can find two different GAN implementations with the goal of generating paintings in the style of the post-impressionist artist Vincent van Gogh. To achieve this goal, we initially constructed a Deep Convolutional GAN (DCGAN), also known as a vanilla GAN. The actual code, the structure, the weights (aftter 500 iterations) together with the output produced by this network can be found in the DCGAN folder.
As a next step, an optimized and more robust type of GAN was constructed, namely a Wasserstein GAN with gradient penalty (WGAN-GP), which on the other hand, is more challenging to implement, but the training process is less prone to failure. The actual code, the structure and the output of this network can all be found in the WGAN-GP folder. The weights were too big to upload to github, but upon request can be obtained from the author.