2CET-GAN: Pixel-Level GAN Model for Human Facial Expression Transfer

TL;DR

This paper introduces 2CET-GAN, an unsupervised pixel-level GAN model that transfers continuous facial expressions without emotion labels, producing diverse, high-quality results and generalizing well to new identities.

Contribution

The paper presents a novel unsupervised CycleGAN- and InfoGAN-based model for continuous facial expression transfer at the pixel level, overcoming limitations of previous methods.

Findings

Generates diverse, high-quality facial expressions

Successfully generalizes to unseen identities

Operates without emotion labels

Abstract

Recent studies have used GAN to transfer expressions between human faces. However, existing models have many flaws: relying on emotion labels, lacking continuous expressions, and failing to capture the expression details. To address these limitations, we propose a novel CycleGAN- and InfoGAN-based network called 2 Cycles Expression Transfer GAN (2CET-GAN), which can learn continuous expression transfer without using emotion labels. The experiment shows our network can generate diverse and high-quality expressions and can generalize to unknown identities. To the best of our knowledge, we are among the first to successfully use an unsupervised approach to disentangle expression representation from identities at the pixel level.