M$^{3}$T2IBench: A Large-Scale Multi-Category, Multi-Instance, Multi-Relation Text-to-Image Benchmark

ICLR 2026 Conference Submission16333 Authors

Keywords: Text-to-Image, Benchmark, Multi-Instance

TL;DR: We propose M T2IBench, a large-scale, multi-category, multi-instance, multi-relation text-to-image benchmark.

Abstract: Text-to-image models are known to struggle with generating images that perfectly align with textual prompts. Several previous studies have focused on evaluating image-text alignment in text-to-image generation. However, these evaluations either address overly simple scenarios, especially overlooking the difficulty of prompts with multiple different instances belonging to the same category, or they introduce metrics that do not correlate well with human evaluation. In this study, we introduce M$^{3}$T2IBench, a large-scale, multi-category, multi-instance, multi-relation along with an object-detection-based evaluation metric, *AlignScore*, which aligns closely with human evaluation. Our findings reveal that current text-to-image models perform poorly on this challenging benchmark. Additionally, we propose the Revise-Then-Enforce approach to enhance image-text alignment. This training-free post-editing method demonstrates improvements in image-text alignment across a broad range of diffusion models.

Primary Area: datasets and benchmarks

Submission Number: 16333