GAN

Repository for GANs (families of generative adversarial network) by Keras (tensorflow)

requires

• Keras (2.3.1)
• Tensorflow (1.14.0)
• Numpy (1.16.4)
• matplotlib (3.1.0)

All the scripts are refering to https://github.com/eriklindernoren/Keras-GAN . I modified them for explaining the function of the models in detail. Full description is shown in qiita website (but sorry in Japanese).

GAN

今さら聞けないGAN（1）　基本構造の理解 on qiita (in Japanese)

DCGAN

今さら聞けないGAN （2）　DCGANによる画像生成 on qiita (in Japanese)

今さら聞けないGAN（3）　潜在変数と生成画像 on qiita (in Japanese)

WGAN-gp

今さら聞けないGAN（4） WGAN on qiita (in Japanese)

今さら聞けないGAN (5) WGAN-gpの実装 on qiita (in Japanese)

Conditional GAN

今さら聞けないGAN（6） Conditional GANの実装 on qiita (in Japanese)

今さらGAN 聞けないGAN（7）conditional GANの生成画像 on qiita (in Japanese)

I hacked latent space flag from {0,0,0,0,...,0} to {0,0,0,1,...,0} so that genarated images are changed to "3".