M$^{3}$T2IBench: A Large-Scale Multi-Category, Multi-Instance, Multi-Relation Text-to-Image Benchmark
This addresses the need for better evaluation and improvement of text-to-image alignment for complex prompts, though it is incremental as it builds on existing evaluation efforts.
The authors tackled the problem of text-to-image models struggling with complex prompts involving multiple instances and relations by introducing M^3T2IBench, a large-scale benchmark with an object-detection-based metric called AlignScore that correlates well with human evaluation, and found that current models perform poorly on it, but proposed a training-free post-editing method that improves alignment across diffusion models.
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 open-source 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. \footnote{Our code and data has been released in supplementary material and will be made publicly available after the paper is accepted.}