CageNet: A Meta-Framework for Learning on Wild Meshes

2025-05-24

AI Summary

Existing triangle mesh learning methods suffer from severely limited generalization on “wild meshes” commonly encountered in real-world scenarios—characterized by multiple disconnected components, non-manifold geometry, and broken connectivity. To address this, we propose the first configurable meta-framework based on caged geometry: it employs a cage mesh as a unified proxy representation and establishes a differentiable functional mapping between the cage and the input mesh via generalized barycentric coordinates, enabling end-to-end learning on arbitrary topologically defective meshes for the first time. The framework integrates automatic cage construction, differentiable geometric transformations, graph neural networks, and adaptive feature projection. Evaluated on wild-mesh segmentation and skinning weight prediction, our method significantly outperforms state-of-the-art approaches, demonstrating superior robustness and accuracy—particularly on non-manifold and multi-component meshes.

Technology Category

• Machine Learning: Learning with Manifolds
• Computer Vision: Representation Learning for Vision
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Learning on triangle meshes has recently proven to be instrumental to a myriad of tasks, from shape classification, to segmentation, to deformation and animation, to mention just a few. While some of these applications are tackled through neural network architectures which are tailored to the application at hand, many others use generic frameworks for triangle meshes where the only customization required is the modification of the input features and the loss function. Our goal in this paper is to broaden the applicability of these generic frameworks to"wild", i.e. meshes in-the-wild which often have multiple components, non-manifold elements, disrupted connectivity, or a combination of these. We propose a configurable meta-framework based on the concept of caged geometry: Given a mesh, a cage is a single component manifold triangle mesh that envelopes it closely. Generalized barycentric coordinates map between functions on the cage, and functions on the mesh, allowing us to learn and test on a variety of data, in different applications. We demonstrate this concept by learning segmentation and skinning weights on difficult data, achieving better performance to state of the art techniques on wild meshes.

Problem

• Extending generic mesh frameworks to handle wild meshes
• Proposing a configurable meta-framework using caged geometry
• Improving performance on segmentation and skinning tasks

Innovation

• Configurable meta-framework for wild meshes
• Uses caged geometry for manifold wrapping
• Generalized barycentric coordinates enable mapping