Domain Re-Modulation for Few-Shot Generative Domain Adaptation

TL;DR

This paper introduces DoRM, a novel generator structure for few-shot generative domain adaptation that mimics human brain learning, enabling high-quality, diverse, and multi-domain image synthesis with improved cross-domain consistency.

Contribution

The paper proposes the Domain Re-Modulation (DoRM) structure with memory and domain association, enhancing few-shot GDA by enabling multi-domain generation and better cross-domain consistency.

Findings

DoRM achieves superior quality and diversity in few-shot GDA.

The similarity-based structure loss improves cross-domain consistency.

The method enables high-fidelity multi-domain and hybrid-domain generation.

Abstract

In this study, we delve into the task of few-shot Generative Domain Adaptation (GDA), which involves transferring a pre-trained generator from one domain to a new domain using only a few reference images. Inspired by the way human brains acquire knowledge in new domains, we present an innovative generator structure called Domain Re-Modulation (DoRM). DoRM not only meets the criteria of high quality, large synthesis diversity, and cross-domain consistency, which were achieved by previous research in GDA, but also incorporates memory and domain association, akin to how human brains operate. Specifically, DoRM freezes the source generator and introduces new mapping and affine modules (M&A modules) to capture the attributes of the target domain during GDA. This process resembles the formation of new synapses in human brains. Consequently, a linearly combinable domain shift occurs in the style space. By incorporating multiple new M&A modules, the generator gains the capability to perform high-fidelity multi-domain and hybrid-domain generation. Moreover, to maintain cross-domain consistency more effectively, we introduce a similarity-based structure loss. This loss aligns the auto-correlation map of the target image with its corresponding auto-correlation map of the source image during training. Through extensive experiments, we demonstrate the superior performance of our DoRM and similarity-based structure loss in few-shot GDA, both quantitatively and qualitatively. The code will be available at https://github.com/wuyi2020/DoRM.