通过奖励建模增强图像生成中的空间理解
Enhancing Spatial Understanding in Image Generation via Reward Modeling
用奖励模型为文生图的空间关系正确性提供可微/可打分信号，可直接接入在线 RL（如 GRPO/PPO 类）微调扩散或自回归生成器，解决复杂空间提示词需多次采样的问题；SpatialScore 也可作为空间一致性的自动评测器复用。
原文依据 this reward model effectively enables online reinforcement learning for the complex spatial generation
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论近期文本到图像生成的进展大幅提升了视觉保真度与创造力，但同时也对提示词的复杂度提出了更高要求——尤其是在编码复杂空间关系方面。在此类情况下，要获得满意结果往往需要多次采样尝试。为应对这一挑战，我们提出一种新方法，以增强现有图像生成模型的空间理解能力。我们首先构建了 SpatialReward-Dataset，包含超过 8 万对偏好数据。基于该数据集，我们构建了 SpatialScore，这是一个用于评估文本到图像生成中空间关系准确性的奖励模型，其性能在空间评估上甚至超越了领先的专有模型。我们进一步证明，该奖励模型能够有效支撑面向复杂空间生成的在线强化学习。在多个基准上的大量实验表明，我们的专用奖励模型为图像生成的空间理解带来了显著且一致的提升。所有模型与数据集都将公开。
英文原文摘要 1106 字符
Recent progress in text-to-image generation has greatly advanced visual fidelity and creativity, but it has also imposed higher demands on prompt complexity--particularly in encoding intricate spatial relationships. In such cases, achieving satisfactory results often requires multiple sampling attempts.To address this challenge, we introduce a novel method that strengthens the spatial understanding of current image generation models. We first construct the SpatialReward-Dataset with over 80k preference pairs. Building on this dataset, we build SpatialScore, a reward model designed to evaluate the accuracy of spatial relationships in text-to-image generation, achieving performance that even surpasses leading proprietary models on spatial evaluation.We further demonstrate that this reward model effectively enables online reinforcement learning for the complex spatial generation. Extensive experiments across multiple benchmarks show that our specialized reward model yields significant and consistent gains in spatial understanding for image generation. All models and datasets will be released.