常见修复物体的上下文内外数据集
Common Inpainted Objects In-N-Out of Context
用扩散修复在 COCO 上系统性地替换物体，构造出上下文一致与不一致的受控样本，为上下文推理、上下文预测和伪造检测提供了可复用的评测床；做图像取证或上下文理解的研究者可以直接用这套数据做零微调评测。
原文依据 By systematically replacing objects in COCO images through diffusion-based inpainting, we create 97,722 unique images
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论我们提出常见修复物体的上下文内外数据集（COinCO），这是一个旨在解决现有视觉数据集中上下文外样本稀缺问题的新数据集。通过扩散修复系统性地替换 COCO 图像中的物体，我们构建了 97,722 张独特图像，其中既包含上下文连贯的场景，也包含上下文不一致的场景，从而支持有效的上下文学习。每个被修复的物体都经过细致核验，并借助大型视觉语言模型的评估被归类为上下文内或上下文外。我们展示了 COinCO 所支撑的三项关键任务：(1) 一种细粒度上下文推理方法，依据三项准则将物体分类为上下文内或上下文外；(2) 一项新颖的「由上下文预测物体」任务，在实例级与团簇级语义上判定哪些新物体自然地属于给定场景；(3) 在最优方法上无需微调的上下文增强伪造检测。COinCO 提供了一个带有上下文变化的受控测试平台，为推进计算机视觉中上下文感知的视觉理解（包括图像取证）奠定了基础。代码与数据集见 https://co-in-co.github.io/。
英文原文摘要 1185 字符
We present Common Inpainted Objects In-N-Out of Context (COinCO), a novel dataset addressing the scarcity of out-of-context examples in existing vision datasets. By systematically replacing objects in COCO images through diffusion-based inpainting, we create 97,722 unique images featuring both contextually coherent and inconsistent scenes, enabling effective context learning. Each inpainted object is meticulously verified and categorized as in- or out-of-context through Large Vision Language Model assessments. We demonstrate three key tasks enabled by COinCO: (1) a fine-grained context reasoning approach that classifies objects as in- or out-of-context based on three criteria; (2) a novel Objects-from-Context prediction task that determines which new objects naturally belong in given scenes at both instance and clique level semantics, and (3) context-enhanced fake detection on state-of-the-art methods without fine-tuning. COinCO provides a controlled testbed with contextual variations, establishing a foundation for advancing context-aware visual understanding in computer vision, including image forensics. Code and dataset are available at https://co-in-co.github.io/.