CCCaption: 为完整和正确图像描述的双奖励强化学习
CCCaption: Dual-Reward Reinforcement Learning for Complete and Correct Image Captioning
AI总结：
本文提出CCCaption框架，通过双奖励强化学习优化图像描述的完整性和正确性，利用多样化LVLMs和动态查询采样策略提升训练效率，并通过验证子描述查询的真实性来惩罚幻觉，从而生成更符合客观标准的描述。
摘要：
图像描述仍然是视觉语言理解的基础任务，但地面真实监督仍主要依赖人工标注的参考。由于人工标注反映主观偏好和专业知识，地面真实描述往往不完整甚至不正确，从而限制了描述模型。我们主张通过两个客观方面评估描述质量：完整性（描述是否涵盖所有显著的视觉事实？）和正确性（描述是否与图像真实相符？）。为此，我们引入CCCaption：一个双奖励强化学习框架，配有专门的微调语料库，明确优化这些属性以生成完整且正确的描述。为了完整性，我们使用多样化的LVLMs将图像分解为一组视觉查询，并奖励能够回答更多查询的描述，采用动态查询采样策略以提高训练效率。为了正确性，我们通过验证子描述查询的真实性来惩罚包含幻觉的描述，这些子描述查询来源于描述分解。我们的对称双奖励优化共同最大化完整性和正确性，引导模型生成更符合这些客观标准的描述。在标准描述任务基准上的广泛实验显示了一致的改进，为训练超越人工标注模仿的描述模型提供了系统性的路径。
英文摘要：
Image captioning remains a fundamental task for vision language understanding, yet ground-truth supervision still relies predominantly on human-annotated references. Because human annotations reflect subjective preferences and expertise, ground-truth captions are often incomplete or even incorrect, which in turn limits caption models. We argue that caption quality should be assessed by two objective aspects: completeness (does the caption cover all salient visual facts?) and correctness (are the descriptions true with respect to the image?). To this end, we introduce CCCaption: a dual-reward reinforcement learning framework with a dedicated fine-tuning corpus that explicitly optimizes these properties to generate \textbf{C}omplete and \textbf{C}orrect \textbf{Captions}. For completeness, we use diverse LVLMs to disentangle the image into a set of visual queries, and reward captions that answer more of these queries, with a dynamic query sampling strategy to improve training efficiency. For correctness, we penalize captions that contain hallucinations by validating the authenticity of sub-caption queries, which are derived from the caption decomposition. Our symmetric dual-reward optimization jointly maximizes completeness and correctness, guiding models toward captions that better satisfy these objective criteria. Extensive experiments across standard captioning benchmarks show consistent improvements, offering a principled path to training caption models beyond human-annotation imitation.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
