AnimeGAN-Keras

A Collection of my reimplemented GAN architectures which are my most favorite machine learning models.

1. Vanila GAN
2. Wasserstein GAN with Gradient Penalty
3. StyleGAN without progressive training

Datasets

Those GAN models were trained using soumik12345's dataset and Danbooru dataset processed by GWERN

soumik12345's dataset: https://www.kaggle.com/soumikrakshit/anime-faces
Danbooru dataset from GWERN: https://www.gwern.net/TWDNE

Samples from soumik12345's dataset

Generated samples from DCGAN

Generated samples from StyleGAN

Btw, I also like other GAN but not yet know how to implement xD

References

Goodfellow, I. J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., … Bengio, Y. (2014). Generative Adversarial Networks. doi:10.48550/ARXIV.1406.2661

Arjovsky, M., Chintala, S. & Bottou, L.. (2017). Wasserstein Generative Adversarial Networks. Proceedings of the 34th International Conference on Machine Learning, in Proceedings of Machine Learning Research 70:214-223 Available from https://proceedings.mlr.press/v70/arjovsky17a.html.

Karras, T., Laine, S., & Aila, T. (2018). A Style-Based Generator Architecture for Generative Adversarial Networks. doi:10.48550/ARXIV.1812.04948

Karras, T., Laine, S., Aittala, M., Hellsten, J., Lehtinen, J., & Aila, T. (2019). Analyzing and Improving the Image Quality of StyleGAN. doi:10.48550/ARXIV.1912.04958