One-shot Unsupervised Domain Adaptation with Personalized Diffusion Models

TL;DR

This paper introduces DATUM, a novel method using text-guided diffusion models to generate synthetic target domain images for one-shot unsupervised domain adaptation, significantly improving segmentation performance.

Contribution

The paper proposes a new approach leveraging diffusion models and text guidance to generate diverse, semantically controlled images for OSUDA, surpassing existing style transfer methods.

Findings

Outperforms state-of-the-art OSUDA methods by up to +7.1%.

Generates photo-realistic, semantically guided images for target domain adaptation.

Enables flexible, context-preserving image synthesis from a single unlabeled sample.

Abstract

Adapting a segmentation model from a labeled source domain to a target domain, where a single unlabeled datum is available, is one the most challenging problems in domain adaptation and is otherwise known as one-shot unsupervised domain adaptation (OSUDA). Most of the prior works have addressed the problem by relying on style transfer techniques, where the source images are stylized to have the appearance of the target domain. Departing from the common notion of transferring only the target ``texture'' information, we leverage text-to-image diffusion models (e.g., Stable Diffusion) to generate a synthetic target dataset with photo-realistic images that not only faithfully depict the style of the target domain, but are also characterized by novel scenes in diverse contexts. The text interface in our method Data AugmenTation with diffUsion Models (DATUM) endows us with the possibility of guiding the generation of images towards desired semantic concepts while respecting the original spatial context of a single training image, which is not possible in existing OSUDA methods. Extensive experiments on standard benchmarks show that our DATUM surpasses the state-of-the-art OSUDA methods by up to +7.1%. The implementation is available at https://github.com/yasserben/DATUM