用度量认知地图实现3D视觉模型的可解释空间推理
Map2Thought: Explicit 3D Spatial Reasoning via Metric Cognitive Maps
- 构建度量认知地图融合离散关系与连续几何表示
- 仅用一半标注数据达59.9%准确率，接近全量训练60.9%水平
- 在小样本场景下显著优于当前最优方法，适合可解释性需求强的场景
我们提出Map2Thought框架，为3D视觉语言模型提供显式且可解释的空间推理能力。该框架基于两个核心组件：度量认知地图（Metric-CogMap）和认知思维链（Cog-CoT）。Metric-CogMap通过融合离散网格的关系推理能力与连续度量尺度的几何理解能力，构建统一的空间表征。在此基础上，Cog-CoT通过确定性操作（如向量运算、边界框距离计算、遮挡感知的外观顺序线索）进行显式几何推理，生成基于3D结构的可解释推理轨迹。实验表明，Map2Thought实现了可解释的3D理解，在仅使用一半监督信号的情况下达到59.9%的准确率，接近全数据集训练的60.9%基准表现。在VSI-Bench数据集上，于10%、25%、50%训练子集下，分别领先当前最优方法5.3%、4.8%、4.0%，展现出卓越的小样本泛化能力。
原文摘要 · Abstract (English)
We propose Map2Thought, a framework that enables explicit and interpretable spatial reasoning for 3D VLMs. The framework is grounded in two key components: Metric Cognitive Map (Metric-CogMap) and Cognitive Chain-of-Thought (Cog-CoT). Metric-CogMap provides a unified spatial representation by integrating a discrete grid for relational reasoning with a continuous, metric-scale representation for precise geometric understanding. Building upon the Metric-CogMap, Cog-CoT performs explicit geometric reasoning through deterministic operations, including vector operations, bounding-box distances, and occlusion-aware appearance order cues, producing interpretable inference traces grounded in 3D structure. Experimental results show that Map2Thought enables explainable 3D understanding, achieving 59.9% accuracy using only half the supervision, closely matching the 60.9% baseline trained with the full dataset. It consistently outperforms state-of-the-art methods by 5.3%, 4.8%, and 4.0% under 10%, 25%, and 50% training subsets, respectively, on the VSI-Bench.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。