用特征场学习实现3D形状的高质量凸分解，支持开放世界通用场景。
Learning Convex Decomposition via Feature Fields
- 通过学习连续特征场，基于几何定义自监督训练模型。
- 在网格、CAD和高斯点云上均实现优于现有方法的分解质量。
- 首个无需标注的开放世界凸分解模型，适用于大规模数据训练。
本文提出一种基于特征场学习的新方法，解决长期存在的凸分解难题，首次实现开集环境下3D形状的前馈式凸分解。该方法将3D形状分解为凸体的并集，对物理仿真中的碰撞检测等任务至关重要。核心思想是采用特征学习策略，构建连续特征场，并通过从凸性经典定义推导出的自监督纯几何目标进行聚类，获得优质分解结果。该方法既可用于单个形状优化，更关键的是可实现大规模数据上的可扩展自监督学习，从而建立首个已学习的开集凸分解模型。实验表明，该方法在网格、CAD模型及高斯点云等多种表示形式下，分解质量优于现有方法，且具备良好的跨对象与跨表示泛化能力。
原文摘要 · Abstract (English)
This work proposes a new formulation to the long-standing problem of convex decomposition through learning feature fields, enabling the first feed-forward model for open-world convex decomposition. Our method produces high-quality decompositions of 3D shapes into a union of convex bodies, which are essential to accelerate collision detection in physical simulation, amongst many other applications. The key insight is to adopt a feature learning approach and learn a continuous feature field that can later be clustered to yield a good convex decomposition via our self-supervised, purely-geometric objective derived from the classical definition of convexity. Our formulation can be used for single shape optimization, but more importantly, feature prediction unlocks scalable, self-supervised learning on large datasets resulting in the first learned open-world model for convex decomposition. Experiments show that our decompositions are higher-quality than alternatives and generalize across open-world objects as well as across representations to meshes, CAD models, and even Gaussian splats. https://research.nvidia.com/labs/sil/projects/learning-convex-decomp/
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。