Large Scale GAN Training for High Fidelity Natural Image Synthesis

09/28/2018

∙

Despite recent progress in generative image modeling, successfully generating high-resolution, diverse samples from complex datasets such as ImageNet remains an elusive goal. To this end, we train Generative Adversarial Networks at the largest scale yet attempted, and study the instabilities specific to such scale. We find that applying orthogonal regularization to the generator renders it amenable to a simple "truncation trick", allowing fine control over the trade-off between sample fidelity and variety by truncating the latent space. Our modifications lead to models which set the new state of the art in class-conditional image synthesis. When trained on ImageNet at 128x128 resolution, our models (BigGANs) achieve an Inception Score (IS) of 166.3 and Frechet Inception Distance (FID) of 9.6, improving over the previous best IS of 52.52 and FID of 18.65.

READ FULL TEXT

Large Scale GAN Training for High Fidelity Natural Image Synthesis

Efficient Video Generation on Complex Datasets

High-Fidelity Image Generation With Fewer Labels

Improving the Speed and Quality of GAN by Adversarial Training

StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasets

Make a Face: Towards Arbitrary High Fidelity Face Manipulation

Group Equivariant Generative Adversarial Networks

CHIMLE: Conditional Hierarchical IMLE for Multimodal Conditional Image Synthesis

Large Scale GAN Training for High Fidelity Natural Image Synthesis

Related Research

Efficient Video Generation on Complex Datasets

High-Fidelity Image Generation With Fewer Labels

Improving the Speed and Quality of GAN by Adversarial Training

StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasets

Make a Face: Towards Arbitrary High Fidelity Face Manipulation

Group Equivariant Generative Adversarial Networks

CHIMLE: Conditional Hierarchical IMLE for Multimodal Conditional Image Synthesis