从条目化文本监督中学习完整且可解释的视觉表征
Learning complete and explainable visual representations from itemized text supervision
针对单图多条目文本（如医学影像的多处独立发现）的监督，用交叉注意力生成条目条件化的视觉嵌入，并设计条目独立性与覆盖完整性目标，可复用于非物体中心领域的视觉-语言预训练与细粒度定位式解释。
原文依据 ItemizedCLIP employs a cross-attention module to produce text item-conditioned visual embeddings
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论以语言监督训练视觉模型能够获得通用且可迁移的表征。然而，许多视觉领域，尤其是医学影像与遥感等非以物体为中心的领域，包含条目化的文本标注：多个文本条目描述同一图像中彼此不同且语义独立的发现。这类监督不同于标准的多描述监督，后者中的描述彼此冗余或高度重叠。为此，我们提出 ItemizedCLIP，一个从条目化文本监督中学习完整且可解释视觉表征的框架。ItemizedCLIP 采用交叉注意力模块生成以文本条目为条件的视觉嵌入，并设计了一组定制目标，联合约束条目独立性（不同条目对应不同区域）与表征完整性（覆盖所有条目）。在四个天然具有条目化文本监督的领域（脑部 MRI、头部 CT、胸部 CT、遥感）以及一个额外的人工合成条目化数据集上，ItemizedCLIP 在零样本性能与细粒度可解释性上相较基线取得显著提升。所得 ItemizedCLIP 表征具有语义接地、条目可区分、完整且视觉可解释的特点。
英文原文摘要 1276 字符
Training vision models with language supervision enables general and transferable representations. However, many visual domains, especially non-object-centric domains such as medical imaging and remote sensing, contain itemized text annotations: multiple text items describing distinct and semantically independent findings within a single image. Such supervision differs from standard multi-caption supervision, where captions are redundant or highly overlapping. Here, we introduce ItemizedCLIP, a framework for learning complete and explainable visual representations from itemized text supervision. ItemizedCLIP employs a cross-attention module to produce text item-conditioned visual embeddings and a set of tailored objectives that jointly enforce item independence (distinct regions for distinct items) and representation completeness (coverage of all items). Across four domains with naturally itemized text supervision (brain MRI, head CT, chest CT, remote sensing) and one additional synthetically itemized dataset, ItemizedCLIP achieves substantial improvements in zero-shot performance and fine-grained interpretability over baselines. The resulting ItemizedCLIP representations are semantically grounded, item-differentiable, complete, and visually interpretable.