Saved in:
Bibliographic Details
Main Authors: An, Jingkun, Zhu, Yinghao, Li, Zongjian, Zhou, Enshen, Feng, Haoran, Huang, Xijie, Chen, Bohua, Shi, Yemin, Pan, Chengwei
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2403.13352
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913630507761664
author An, Jingkun
Zhu, Yinghao
Li, Zongjian
Zhou, Enshen
Feng, Haoran
Huang, Xijie
Chen, Bohua
Shi, Yemin
Pan, Chengwei
author_facet An, Jingkun
Zhu, Yinghao
Li, Zongjian
Zhou, Enshen
Feng, Haoran
Huang, Xijie
Chen, Bohua
Shi, Yemin
Pan, Chengwei
contents Text-to-Image (T2I) diffusion models have achieved remarkable success in image generation. Despite their progress, challenges remain in both prompt-following ability, image quality and lack of high-quality datasets, which are essential for refining these models. As acquiring labeled data is costly, we introduce AGFSync, a framework that enhances T2I diffusion models through Direct Preference Optimization (DPO) in a fully AI-driven approach. AGFSync utilizes Vision-Language Models (VLM) to assess image quality across style, coherence, and aesthetics, generating feedback data within an AI-driven loop. By applying AGFSync to leading T2I models such as SD v1.4, v1.5, and SDXL-base, our extensive experiments on the TIFA dataset demonstrate notable improvements in VQA scores, aesthetic evaluations, and performance on the HPSv2 benchmark, consistently outperforming the base models. AGFSync's method of refining T2I diffusion models paves the way for scalable alignment techniques. Our code and dataset are publicly available at https://anjingkun.github.io/AGFSync.
format Preprint
id arxiv_https___arxiv_org_abs_2403_13352
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle AGFSync: Leveraging AI-Generated Feedback for Preference Optimization in Text-to-Image Generation
An, Jingkun
Zhu, Yinghao
Li, Zongjian
Zhou, Enshen
Feng, Haoran
Huang, Xijie
Chen, Bohua
Shi, Yemin
Pan, Chengwei
Computer Vision and Pattern Recognition
Text-to-Image (T2I) diffusion models have achieved remarkable success in image generation. Despite their progress, challenges remain in both prompt-following ability, image quality and lack of high-quality datasets, which are essential for refining these models. As acquiring labeled data is costly, we introduce AGFSync, a framework that enhances T2I diffusion models through Direct Preference Optimization (DPO) in a fully AI-driven approach. AGFSync utilizes Vision-Language Models (VLM) to assess image quality across style, coherence, and aesthetics, generating feedback data within an AI-driven loop. By applying AGFSync to leading T2I models such as SD v1.4, v1.5, and SDXL-base, our extensive experiments on the TIFA dataset demonstrate notable improvements in VQA scores, aesthetic evaluations, and performance on the HPSv2 benchmark, consistently outperforming the base models. AGFSync's method of refining T2I diffusion models paves the way for scalable alignment techniques. Our code and dataset are publicly available at https://anjingkun.github.io/AGFSync.
title AGFSync: Leveraging AI-Generated Feedback for Preference Optimization in Text-to-Image Generation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2403.13352