Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox

Diverse Image Generation via Self-Conditioned GANs

Jun 18, 2020
Steven Liu, Tongzhou Wang, David Bau, Jun-Yan Zhu, Antonio Torralba

Share this with someone who'll enjoy it:

We introduce a simple but effective unsupervised method for generating realistic and diverse images. We train a class-conditional GAN model without using manually annotated class labels. Instead, our model is conditional on labels automatically derived from clustering in the discriminator's feature space. Our clustering step automatically discovers diverse modes, and explicitly requires the generator to cover them. Experiments on standard mode collapse benchmarks show that our method outperforms several competing methods when addressing mode collapse. Our method also performs well on large-scale datasets such as ImageNet and Places365, improving both image diversity and standard quality metrics, compared to previous methods.

* CVPR 2020. Code: Webpage: 

   Access Paper Source

Share this with someone who'll enjoy it: