DCGAN_Keras

Tensorflow (Keras) implementation of Deep Convolutional Generative Adversarial Networks.

Approach

This is an implementation of DCGAN on CelebA dataset. I have sampled 10,000 images from the whole dataset and trained the model for 10,000 iterations.

Model

Model architecture for Generator, Discriminator, and GAN:

Generator

Discriminator

GAN

Results

For results and weights, refer to this.