T2I-Eval-R1: Reinforcement Learning-Driven Reasoning for Interpretable Text-to-Image Evaluation
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_ | 1866915301309808640 |
|---|---|
| author | Ma, Zi-Ao Lan, Tian Tu, Rong-Cheng Liu, Shu-Hang Huang, Heyan Wu, Zhijing Xu, Chen Mao, Xian-Ling |
| author_facet | Ma, Zi-Ao Lan, Tian Tu, Rong-Cheng Liu, Shu-Hang Huang, Heyan Wu, Zhijing Xu, Chen Mao, Xian-Ling |
| contents | The rapid progress in diffusion-based text-to-image (T2I) generation has created an urgent need for interpretable automatic evaluation methods that can assess the quality of generated images, therefore reducing the human annotation burden. To reduce the prohibitive cost of relying on commercial models for large-scale evaluation, and to improve the reasoning capabilities of open-source models, recent research has explored supervised fine-tuning (SFT) of multimodal large language models (MLLMs) as dedicated T2I evaluators. However, SFT approaches typically rely on high-quality critique datasets, which are either generated by proprietary LLMs-with potential issues of bias and inconsistency-or annotated by humans at high cost, limiting their scalability and generalization. To address these limitations, we propose T2I-Eval-R1, a novel reinforcement learning framework that trains open-source MLLMs using only coarse-grained quality scores, thereby avoiding the need for annotating high-quality interpretable evaluation rationale. Our approach integrates Group Relative Policy Optimization (GRPO) into the instruction-tuning process, enabling models to generate both scalar scores and interpretable reasoning chains with only easy accessible annotated judgment scores or preferences. Furthermore, we introduce a continuous reward formulation that encourages score diversity and provides stable optimization signals, leading to more robust and discriminative evaluation behavior. Experimental results on three established T2I meta-evaluation benchmarks demonstrate that T2I-Eval-R1 achieves significantly higher alignment with human assessments and offers more accurate interpretable score rationales compared to strong baseline methods. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_17897 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | T2I-Eval-R1: Reinforcement Learning-Driven Reasoning for Interpretable Text-to-Image Evaluation Ma, Zi-Ao Lan, Tian Tu, Rong-Cheng Liu, Shu-Hang Huang, Heyan Wu, Zhijing Xu, Chen Mao, Xian-Ling Artificial Intelligence Computation and Language The rapid progress in diffusion-based text-to-image (T2I) generation has created an urgent need for interpretable automatic evaluation methods that can assess the quality of generated images, therefore reducing the human annotation burden. To reduce the prohibitive cost of relying on commercial models for large-scale evaluation, and to improve the reasoning capabilities of open-source models, recent research has explored supervised fine-tuning (SFT) of multimodal large language models (MLLMs) as dedicated T2I evaluators. However, SFT approaches typically rely on high-quality critique datasets, which are either generated by proprietary LLMs-with potential issues of bias and inconsistency-or annotated by humans at high cost, limiting their scalability and generalization. To address these limitations, we propose T2I-Eval-R1, a novel reinforcement learning framework that trains open-source MLLMs using only coarse-grained quality scores, thereby avoiding the need for annotating high-quality interpretable evaluation rationale. Our approach integrates Group Relative Policy Optimization (GRPO) into the instruction-tuning process, enabling models to generate both scalar scores and interpretable reasoning chains with only easy accessible annotated judgment scores or preferences. Furthermore, we introduce a continuous reward formulation that encourages score diversity and provides stable optimization signals, leading to more robust and discriminative evaluation behavior. Experimental results on three established T2I meta-evaluation benchmarks demonstrate that T2I-Eval-R1 achieves significantly higher alignment with human assessments and offers more accurate interpretable score rationales compared to strong baseline methods. |
| title | T2I-Eval-R1: Reinforcement Learning-Driven Reasoning for Interpretable Text-to-Image Evaluation |
| topic | Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2505.17897 |