Text-to-image models often struggle to generate images that precisely match textual prompts. Prior research has extensively studied the evaluation of image-text alignment in text-to-image generation. However, existing evaluations primarily focus on agreement with human assessments, neglecting other critical properties of a trustworthy evaluation framework. In this work, we first identify two key aspects that a reliable evaluation should address. We then empirically demonstrate that current mainstream evaluation frameworks fail to fully satisfy these properties across a diverse range of metrics and models. Finally, we propose recommendations for improving image-text alignment evaluation.
View on arXiv@article{zhang2025_2506.08480,
title={ Re-Thinking the Automatic Evaluation of Image-Text Alignment in Text-to-Image Models },
author={ Huixuan Zhang and Xiaojun Wan },
journal={arXiv preprint arXiv:2506.08480},
year={ 2025 }
}