Source-Free Domain Adaptation with Diffusion-Guided Source Data Generation

TL;DR

This paper presents a novel source-free domain adaptation method using diffusion models to generate source-like images guided by target features, improving adaptation performance across multiple datasets.

Contribution

The paper introduces a diffusion-guided source data generation technique for source-free domain adaptation, leveraging pre-trained diffusion models and a mixup strategy to reduce domain gap.

Findings

Significant performance improvements on Office-31, Office-Home, and VisDA datasets.

Effective generation of source-like images that enhance domain adaptation.

Demonstrated potential of diffusion models in domain adaptation tasks.

Abstract

This paper introduces a novel approach to leverage the generalizability of Diffusion Models for Source-Free Domain Adaptation (DM-SFDA). Our proposed DMSFDA method involves fine-tuning a pre-trained text-to-image diffusion model to generate source domain images using features from the target images to guide the diffusion process. Specifically, the pre-trained diffusion model is fine-tuned to generate source samples that minimize entropy and maximize confidence for the pre-trained source model. We then use a diffusion model-based image mixup strategy to bridge the domain gap between the source and target domains. We validate our approach through comprehensive experiments across a range of datasets, including Office-31, Office-Home, and VisDA. The results demonstrate significant improvements in SFDA performance, highlighting the potential of diffusion models in generating contextually relevant, domain-specific images.