A Unified Feature Disentangler for Multi-Domain Image Translation and Manipulation

TL;DR

This paper introduces a unified deep learning framework that learns domain-invariant features for multi-domain image translation, manipulation, and unsupervised domain adaptation, enabling continuous cross-domain transformations.

Contribution

The proposed model uniquely combines adversarial training with domain-specific information to achieve continuous image translation and effective feature disentanglement across multiple domains.

Findings

Effective in continuous cross-domain image translation

Produces high-quality manipulated images

Enhances unsupervised domain adaptation performance

Abstract

We present a novel and unified deep learning framework which is capable of learning domain-invariant representation from data across multiple domains. Realized by adversarial training with additional ability to exploit domain-specific information, the proposed network is able to perform continuous cross-domain image translation and manipulation, and produces desirable output images accordingly. In addition, the resulting feature representation exhibits superior performance of unsupervised domain adaptation, which also verifies the effectiveness of the proposed model in learning disentangled features for describing cross-domain data.