DomainGallery: Few-shot Domain-driven Image Generation by Attribute-centric Finetuning

TL;DR

DomainGallery is a novel method that fine-tunes pretrained image generation models on few-shot datasets using attribute-centric techniques, enabling effective domain-specific image synthesis.

Contribution

It introduces an attribute-centric finetuning approach with prior attribute erasure, disentanglement, regularization, and enhancement for few-shot domain-driven image generation.

Findings

Outperforms previous methods in domain-specific image generation

Effective with limited training data in various domains

Demonstrates superior quality and diversity in generated images

Abstract

The recent progress in text-to-image models pretrained on large-scale datasets has enabled us to generate various images as long as we provide a text prompt describing what we want. Nevertheless, the availability of these models is still limited when we expect to generate images that fall into a specific domain either hard to describe or just unseen to the models. In this work, we propose DomainGallery, a few-shot domain-driven image generation method which aims at finetuning pretrained Stable Diffusion on few-shot target datasets in an attribute-centric manner. Specifically, DomainGallery features prior attribute erasure, attribute disentanglement, regularization and enhancement. These techniques are tailored to few-shot domain-driven generation in order to solve key issues that previous works have failed to settle. Extensive experiments are given to validate the superior performance of DomainGallery on a variety of domain-driven generation scenarios. Codes are available at https://github.com/Ldhlwh/DomainGallery.