MeT: A Graph Transformer for Semantic Segmentation of 3D Meshes

TL;DR

This paper introduces MeT, a transformer-based approach for 3D mesh semantic segmentation that leverages graph-aware positional encoding and clustering features, achieving state-of-the-art results across multiple datasets.

Contribution

The paper proposes a novel transformer architecture for 3D mesh segmentation using Laplacian eigenvector positional encoding and clustering features, improving modeling of mesh structure.

Findings

Achieves state-of-the-art accuracy on multiple 3D mesh segmentation datasets.

Outperforms CNN-based methods in semantic segmentation tasks.

Demonstrates the effectiveness of graph-aware positional encoding in transformers.

Abstract

Polygonal meshes have become the standard for discretely approximating 3D shapes, thanks to their efficiency and high flexibility in capturing non-uniform shapes. This non-uniformity, however, leads to irregularity in the mesh structure, making tasks like segmentation of 3D meshes particularly challenging. Semantic segmentation of 3D mesh has been typically addressed through CNN-based approaches, leading to good accuracy. Recently, transformers have gained enough momentum both in NLP and computer vision fields, achieving performance at least on par with CNN models, supporting the long-sought architecture universalism. Following this trend, we propose a transformer-based method for semantic segmentation of 3D mesh motivated by a better modeling of the graph structure of meshes, by means of global attention mechanisms. In order to address the limitations of standard transformer architectures in modeling relative positions of non-sequential data, as in the case of 3D meshes, as well as in capturing the local context, we perform positional encoding by means the Laplacian eigenvectors of the adjacency matrix, replacing the traditional sinusoidal positional encodings, and by introducing clustering-based features into the self-attention and cross-attention operators. Experimental results, carried out on three sets of the Shape COSEG Dataset, on the human segmentation dataset proposed in Maron et al., 2017 and on the ShapeNet benchmark, show how the proposed approach yields state-of-the-art performance on semantic segmentation of 3D meshes.