探测并桥接视觉基础模型中可供性推理的几何-交互线索
Probing and Bridging Geometry-Interaction Cues for Affordance Reasoning in Vision Foundation Models
对做视觉研究的人，这篇论文提供了一种无需训练、零样本的可供性估计思路：把 DINO 的部件级几何原型与 Flux 的动词条件空间注意力图直接融合，可作为视觉基础模型内部表征探测与多模态先验组合的通用手段。
原文依据 By simply fusing DINO's geometric prototypes with Flux's interaction maps in a training-free, zero-shot manner
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论一个视觉系统真正理解可供性意味着什么？我们认为，这种理解取决于两种互补的能力：几何感知，即识别物体中使交互得以发生的结构部件；以及交互感知，即建模智能体的动作如何与这些部件发生作用。为检验这一假设，我们对视觉基础模型（VFM）进行了系统性探测。我们发现，DINO 等模型本身编码了部件级的几何结构，而 Flux 等生成模型则包含丰富的、以动词为条件的空间注意力图，可作为隐式的交互先验。关键在于，我们证明这两个维度不仅相互关联，而且是可供性的可组合要素。仅以无需训练、零样本的方式将 DINO 的几何原型与 Flux 的交互图相融合，我们便取得了与弱监督方法相当的可供性估计效果。这一最终的融合实验证实，几何感知与交互感知是 VFM 中可供性理解的基本构件，为感知如何为行动提供依据给出了一种机制性解释。
英文原文摘要 1161 字符
What does it mean for a visual system to truly understand affordance? We argue that this understanding hinges on two complementary capacities: geometric perception, which identifies the structural parts of objects that enable interaction, and interaction perception, which models how an agent's actions engage with those parts. To test this hypothesis, we conduct a systematic probing of Visual Foundation Models (VFMs). We find that models like DINO inherently encode part-level geometric structures, while generative models like Flux contain rich, verb-conditioned spatial attention maps that serve as implicit interaction priors. Crucially, we demonstrate that these two dimensions are not merely correlated but are composable elements of affordance. By simply fusing DINO's geometric prototypes with Flux's interaction maps in a training-free, zero-shot manner, we achieve affordance estimation competitive with weakly-supervised methods. This final fusion experiment confirms that geometric and interaction perception are the fundamental building blocks of affordance understanding in VFMs, providing a mechanistic account of how perception grounds action.