BareBones: 视觉语言模型中零样本几何理解的基准测试
BareBones: Benchmarking Zero-Shot Geometric Comprehension in VLMs
AI总结：
提出BareBones基准，通过去除RGB纹理仅保留轮廓，测试26个视觉语言模型在零样本几何形状理解上的表现，发现模型存在严重的纹理偏置悬崖。
摘要：
尽管视觉语言模型（VLM）在多种多模态任务中展现出卓越的零样本识别能力，但这些架构是否真正理解几何结构，还是仅仅利用RGB纹理和上下文先验作为统计捷径，仍然是一个未解之谜。现有的评估未能分离这一机制，将语义推理与纹理映射混为一谈，并依赖于不精确的标注，这些标注无意中泄露了环境线索。为解决这一空白，我们引入了$ extbf{BareBones}$，一个旨在压力测试纯几何形状理解的零样本基准。我们整理了六个数据集中几何不同类别的像素级轮廓：五个已建立的分割来源（ImageNet-S、DIS5K、ThinObject5K、PASCAL VOC、CUB-200）以及我们新颖的旗舰集合WTP-Bench，建立了一个无噪声的几何分类体系。WTP-Bench是一个极端的、细粒度的视觉谜题，迫使模型仅从边界轮廓中识别类间几何概念。我们对26个最先进的专有和开源权重VLM（例如GPT-4.1、Gemini、Claude Sonnet 4.5、LLaVA）的评估揭示，在去除RGB纹理的情况下，模型性能一致且严重崩溃，我们将这一现象称为$ extit{纹理偏置悬崖}$。通过记录普遍的结构性盲点，BareBones为真正的几何基础建立了一个严格的衡量标准。项目页面：https://eternal-f1ame.github.io/WTP-Bench/
英文摘要：
While Vision-Language Models (VLMs) demonstrate remarkable zero-shot recognition capabilities across a diverse spectrum of multimodal tasks, it yet remains an open question whether these architectures genuinely comprehend geometric structure or merely exploit RGB textures and contextual priors as statistical shortcuts. Existing evaluations fail to isolate this mechanism, conflating semantic reasoning with texture mapping and relying on imprecise annotations that inadvertently leak environmental cues. To address this gap, we introduce $\textbf{BareBones}$, a zero-shot benchmark designed to stress-test pure geometric shape comprehension. We curate pixel-level silhouettes of geometrically distinct classes across six datasets: five established segmentation sources (ImageNet-S, DIS5K, ThinObject5K, PASCAL VOC, CUB-200) and our novel flagship collection, WTP-Bench, establishing a noise-free geometric taxonomy. WTP-Bench is an extreme, fine-grained visual puzzle that forces models to identify inter-class geometric concepts from boundary contours alone. Our evaluation of 26 state-of-the-art proprietary and open-weight VLMs (eg. GPT-4.1, Gemini, Claude Sonnet 4.5, LLaVA) reveals a consistent, severe performance collapse under RGB deprivation, a phenomenon we term the $\textit{Texture Bias Cliff}$. By documenting universal structural blindspots, BareBones establishes a rigorous yardstick for genuine geometric grounding. Project Page: https://eternal-f1ame.github.io/WTP-Bench/
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。
