seonbeomkim / tensorflow-deep-convolutional-gan

(DCGAN)Deep Convolutional Generative Adversarial Network, Conditional-DCGAN

TensorFlow-Deep-Convolutional-GAN

(DCGAN)Deep Convolutional Generative Adversarial Network
(c-DCGAN)Conditional Deep Convolutional Generative Adversarial Network

DCGAN = vanilla GAN + CNN Network + Batch Normalization

DCGAN paper : https://arxiv.org/abs/1511.06434
Batch Normalization paper : https://arxiv.org/abs/1502.03167
Conditional-GAN paper : https://arxiv.org/pdf/1411.1784.pdf

1.DCGAN.py

* 원하는 target이 아닌 랜덤한 이미지 생성
* dataset : MNIST

2.Conditional-DCGAN.py

* DCGAN 모델을 이용하여 원하는 target의 이미지 생성
* dataset : MNIST

DCGAN MNIST result (after 12 epochs of training)

Conditional-DCGAN MNIST result (after 2 epochs of training)