license: mit
language:
- en
ELBO-T2IAlign: A Generic ELBO-Based Method for Calibrating Pixel-level Text-Image Alignment in Diffusion Models
Qin Zhou, Zhiyang Zhang, Jinglong Wang, Xiaobin Li, Jing Zhang*, Qian Yu, Lu Sheng, Dong Xu
Beihang University, University of Hong Kong
Abstract
Diffusion models excel at image generation. Recent studies have shown that these models not only generate high quality images but also encode text-image alignment information through attention maps or loss functions. This information is valuable for various downstream tasks, including segmentation, text-guided image editing, and compositional image generation. However, current methods heavily rely on the assumption of perfect text-image alignment in diffusion models, which is not the case. In this paper, we propose using zero-shot referring image segmentation as a proxy task to evaluate the pixel-level image and class-level text alignment of popular diffusion models. We conduct an in-depth analysis of pixel-text misalignment in diffusion models from the perspective of training data bias. We find that misalignment occurs in images with small sized, occluded, or rare object classes. Therefore, we propose ELBO-T2IAlign—a simple yet effective method to calibrate pixel-text alignment in diffusion models based on the evidence lower bound (ELBO) of likelihood. Our method is training-free and generic, eliminating the need to identify the specific cause of misalignment and works well across various diffusion model architectures. Extensive experiments on commonly used benchmark datasets on image segmentation and generation have verified the effectiveness of our proposed calibration approach.
Details
This repository contains all datasets used in our paper, including COCO, VOC, Context...
Only validation sets are included, which cost about 7G memory.
# unzip command
cat dataset.tar.gz.* > dataset.tar.gz
tar -xzf dataset.tar.gz
Citation
@article{zhou2025elbo,
title={ELBO-T2IAlign: A Generic ELBO-Based Method for Calibrating Pixel-level Text-Image Alignment in Diffusion Models},
author={Zhou, Qin and Zhang, Zhiyang and Wang, Jinglong and Li, Xiaobin and Zhang, Jing and Yu, Qian and Sheng, Lu and Xu, Dong},
journal={arXiv preprint arXiv:2506.09740},
year={2025}
}