通过对话式属性挖掘解决行人检索中的对齐歧义
Tackling Alignment Ambiguity in Person Retrieval through Conversational Attribute Mining
用 MLLM 生成对话式属性描述来补充文本侧细粒度线索，并用双向交叉注意力混合器做 token 级对齐，这一「对话属性挖掘 + token 对齐」思路可迁移到任何细粒度图文检索或跨模态 grounding 任务。
原文依据 we propose CECA, a Conversation-Enhanced Cross-modal Alignment framework
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论文本到图像行人检索（TIPR）旨在根据给定的自然语言描述检索行人图像。由于跨模态对齐中固有的歧义性，该任务仍极具挑战：现有模型往往难以捕捉细粒度对应关系，其对详细行人属性的理解通常局限于局部或粗糙线索，从而导致检索结果错配或错误。为克服这一挑战，我们提出 CECA，一个对话增强的跨模态对齐框架。CECA 通过多模态大语言模型（MLLM）引导的对话强化文本与视觉模态之间的属性对应关系，通过双向交叉注意力混合器（BCM）增强 token 级对齐，并使用置信度感知加权损失（CAWL）降低低质量对话回复的影响以稳定优化。在三个公开基准上的大量实验证明了本方法优越的性能与强大的泛化能力。
英文原文摘要 1042 字符
Text-to-Image Person Retrieval (TIPR) aims to retrieve pedestrian images with a given natural language description. It remains highly challenging due to the inherent ambiguity in cross-modal alignment: existing models often struggle to capture fine-grained correspondences, and their understanding of detailed pedestrian attributes is typically confined to partial or coarse cues, leading to mismatched or erroneous retrieval results. To overcome this challenge, we propose CECA, a Conversation-Enhanced Cross-modal Alignment framework. CECA strengthens the attribute correspondence between textual and visual modalities through multimodal large language models (MLLMs)-guided dialogue, enhances token-level alignment via a Bidirectional Cross-attention Mixer (BCM), and stabilizes optimization with a Confidence-Aware Weighting Loss (CAWL) that reduces the impact of low-quality conversational responses. Extensive experiments on three public benchmarks demonstrate the superior performance and strong generalization ability of our approach.