Re-Thinking the Automatic Evaluation of Image-Text Alignment in Text-to-Image Models
Fuente:
arXiv
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916787742834688 |
|---|---|
| author | Zhang, Huixuan Wan, Xiaojun |
| author_facet | Zhang, Huixuan Wan, Xiaojun |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_08480 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Re-Thinking the Automatic Evaluation of Image-Text Alignment in Text-to-Image Models Zhang, Huixuan Wan, Xiaojun Computation and Language Artificial Intelligence Computer Vision and Pattern Recognition 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. |
| title | Re-Thinking the Automatic Evaluation of Image-Text Alignment in Text-to-Image Models |
| topic | Computation and Language Artificial Intelligence Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2506.08480 |