发表机构
Italian Institute of Technology; University of Genoa; Max Planck Institute for Informatics; University of Bonn; Lamarr Institute; University of Verona（意大利理工学院; 热那亚大学; 马克斯·普朗克信息学研究所; 波恩大学; 拉马尔研究所; 维罗纳大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出基于Transformer的TokenMatch模型，通过曲率引导分词技术学习几何描述符，在多基准测试中实现了更优的3D形状匹配性能，且推理速度更快。
AI 中文摘要
尽管数据驱动的3D形状对应估计近期取得了显著进展，但在部分观测和强非等距变形下的鲁棒匹配仍具挑战性。现有基于学习的方法常依赖手工描述符或基于模板的表示，而近期基于功能映射的生成模型则存在推理成本高、可解释性有限、对部分形状泛化性差的问题。针对这些局限，本文提出TokenMatch，一种基于Transformer的新型统一模型，用于估计3D形状对应。我们的前馈方法仅在BeCoS（一个具有挑战性的非等距部分到部分形状匹配数据集）上训练，无需重新训练或微调即可泛化到完整形状的匹配。TokenMatch利用自注意力和交叉注意力机制，高效学习形状对之间的面片级、点级关系以及密集对应。我们的核心见解是，可利用形状曲率引导将网格自适应分词为面片，从而有效学习用于对应估计的形状特定几何描述符。我们在部分和完整形状匹配的标准基准上评估TokenMatch，包括CP2P、PSMAL、BeCoS、FAUST、SCAPE和SHREC'19。我们的方法始终表现出高性能，在大多数情况下，在平均测地误差和交并比指标上优于现有的部分和完整形状匹配方法，同时推理速度更快，达到亚秒级。
英文摘要
While data-driven 3D shape correspondence estimation has recently seen substantial progress, robust matching under partial observations and strong non-isometric deformations remains challenging. Existing learning-based approaches often rely on hand-crafted descriptors or template-based representations, whereas recent generative models over functional maps suffer from high inference cost, limited interpretability, and poor generalisation to partial shapes. In response to these limitations, this paper introduces TokenMatch, a new transformer-based unified model for estimating 3D shape correspondences. Our feed-forward approach trained exclusively on BeCoS, a challenging non-isometric partial-to-partial shape-matching dataset, can generalise to matching full shapes without retraining or fine-tuning. TokenMatch uses self- and cross-attention mechanisms to efficiently learn patch-level and point-level relations as well as dense correspondences between shape pairs. Our core insight is that meshes can be adaptively tokenised into patches using shape curvature guidance, enabling effective learning of shape-specific geometric descriptors for correspondence estimation. We evaluate TokenMatch on standard benchmarks for partial and full shape matching, including CP2P, PSMAL, BeCoS, FAUST, SCAPE, and SHREC'19. Our method achieves consistently high performance, in most cases outperforming existing methods for partial and full shape matching in the mean geodesic error and intersection-over-union metrics, while also running faster at sub-second inference speeds.
Comments25 pages, 13 figures and 12 tables; project page: https://4dqv.mpi-inf.mpg.de/TokenMatch/
