Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SpatialReward: Verifiable Spatial Reward Modeling for Fine-Grained Spatial Consistency in Text-to-Image Generation

Sashuai Zhou, Qiang Zhou, Junpeng Ma, Yue Cao, Ruofan Hu, Ziang Zhang, Xiaoda Yang, Zhibin Wang, Jun Song, Cheng Yu, Bo Zheng, Zhou Zhao

Abstract

Recent advances in text-to-image (T2I) generation via reinforcement learning (RL) have benefited from reward models that assess semantic alignment and visual quality. However, most existing reward models pay limited attention to fine-grained spatial relationships, often producing images that appear plausible overall yet contain inaccuracies in object positioning. In this work, we present \textbf{SpatialReward}, a verifiable reward model explicitly designed to evaluate spatial layouts in generated images. SpatialReward adopts a multi-stage pipeline: a \emph{Prompt Decomposer} extracts entities, attributes, and spatial metadata from free-form prompts; expert detectors provide accurate visual grounding of object positions and attributes; and a vision-language model applies chain-of-thought reasoning over grounded observations to assess complex spatial relations that are challenging for rule-based methods. To more comprehensively evaluate spatial relationships in generated images, we introduce \textbf{SpatRelBench}, a benchmark covering object attributes, orientation, inter-object relations, and rendered text placement. Experiments on Stable Diffusion and FLUX show that incorporating SpatialReward into RL training consistently improves spatial consistency and overall generation quality, with results aligned more closely to human judgments. These findings indicate that verifiable reward models hold considerable potential for enabling more accurate and controllable optimization in text-to-image generation models.

SpatialReward: Verifiable Spatial Reward Modeling for Fine-Grained Spatial Consistency in Text-to-Image Generation

Abstract

Recent advances in text-to-image (T2I) generation via reinforcement learning (RL) have benefited from reward models that assess semantic alignment and visual quality. However, most existing reward models pay limited attention to fine-grained spatial relationships, often producing images that appear plausible overall yet contain inaccuracies in object positioning. In this work, we present \textbf{SpatialReward}, a verifiable reward model explicitly designed to evaluate spatial layouts in generated images. SpatialReward adopts a multi-stage pipeline: a \emph{Prompt Decomposer} extracts entities, attributes, and spatial metadata from free-form prompts; expert detectors provide accurate visual grounding of object positions and attributes; and a vision-language model applies chain-of-thought reasoning over grounded observations to assess complex spatial relations that are challenging for rule-based methods. To more comprehensively evaluate spatial relationships in generated images, we introduce \textbf{SpatRelBench}, a benchmark covering object attributes, orientation, inter-object relations, and rendered text placement. Experiments on Stable Diffusion and FLUX show that incorporating SpatialReward into RL training consistently improves spatial consistency and overall generation quality, with results aligned more closely to human judgments. These findings indicate that verifiable reward models hold considerable potential for enabling more accurate and controllable optimization in text-to-image generation models.
Paper Structure (25 sections, 12 equations, 5 figures, 4 tables)

This paper contains 25 sections, 12 equations, 5 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. RL-based T2I Optimization
  4. Reward Models for T2I Generation
  5. Benchmarks for T2I Evaluation
  6. SpatialReward
  7. Prompt Decomposition
  8. Fine-grained Verifiable Rewards
  9. Object Attribute and Presence Reward.
  10. Text Content and Localization Reward.
  11. Spatial Chain-of-Thought Reasoning
  12. SpatRelBench
  13. Experiments
  14. Implementation Details
  15. Training Configuration.
  16. ...and 10 more sections

Figures (5)

  • Figure 1: Performance comparison of SD3.5-M sd3.5 optimized via RL using SpatialReward versus Baseline Rewards.
  • Figure 2: Overall framework of our approach. (a) Standard Flow-GRPO flowgrpo reinforcement learning pipeline for text-to-image generation. (b) The proposed SpatialReward, which parses prompts into structured spatial and attribute constraints, verifies them on generated images via expert detection, and uses vision–language chain‑of‑thought reasoning to produce the final reward score.
  • Figure 3: Overview of SpatRelBench, depicting benchmark tasks and their data distribution (a), the construction pipeline (b), and the evaluation methodology (c) designed to assess spatial relation understanding in text‑to‑image models.
  • Figure 4: Qualitative comparison of generated image quality across different methods.
  • Figure 5: Effect of CoT reasoning in spatial relations. CoT combines bounding boxes, orientation, and scene semantics, yielding correct classifications where detected-only matching fails.