My implementation of a DCGAN in keras with tensorflow.

The usage is simple, run gan.py to train the generator and discriminator, if you want to generate images, just execute generate.py. You can create a gif from your training image by executing gif.py.

training image are in the "car" folder, generated images from generate.py are in the "gen" folder (some exemples are already in the folder), the dataset should be in the "car_img" folder (a few images are in just for exemple, if you want the full dataset, please download : http://ai.stanford.edu/~jkrause/cars/car_dataset.html). Models are in the "vroum" folder (vroumdis.h5 for the discriminator and vroumgen.h5 for the generator).