DCGAN-Tensorflow
A Tensorflow implementation of DCGAN proposed in the paper Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks.
Architecture of DCGAN :
About Model:
1. Used strided convolutions (discriminator) and fractional-strided convolutions (generator)
2. BatchNormalization in both the generator and the discriminator network
3. Used LeakyReLU in both generator and discriminator and Tanh in last layer of generator
Results :
| Original | Generated |
 |
 |