Semantic Unfolding of StyleGAN Latent Space

TL;DR

This paper introduces a supervised method using normalizing flows to improve the semantic disentanglement of GAN latent spaces, enhancing face image editing capabilities.

Contribution

It proposes a novel approach to better disentangle facial attributes in GANs by learning a proxy latent space with normalizing flows, addressing limitations of linear attribute separation.

Findings

Enhanced facial attribute disentanglement

More efficient face image editing

Supervised learning improves latent space structure

Abstract

Generative adversarial networks (GANs) have proven to be surprisingly efficient for image editing by inverting and manipulating the latent code corresponding to an input real image. This editing property emerges from the disentangled nature of the latent space. In this paper, we identify that the facial attribute disentanglement is not optimal, thus facial editing relying on linear attribute separation is flawed. We thus propose to improve semantic disentanglement with supervision. Our method consists in learning a proxy latent representation using normalizing flows, and we show that this leads to a more efficient space for face image editing.