Title: Boundary representation learning via Transformer Show Chinese title

Title: 基于Transformer的边界表示学习

Abstract: The recent rise of generative artificial intelligence (AI), powered by Transformer networks, has achieved remarkable success in natural language processing, computer vision, and graphics. However, the application of Transformers in computer-aided design (CAD), particularly for processing boundary representation (B-rep) models, remains largely unexplored. To bridge this gap, this paper introduces Boundary Representation Transformer (BRT), a novel method adapting Transformer for B-rep learning. B-rep models pose unique challenges due to their irregular topology and continuous geometric definitions, which are fundamentally different from the structured and discrete data Transformers are designed for. To address this, BRT proposes a continuous geometric embedding method that encodes B-rep surfaces (trimmed and untrimmed) into B\'ezier triangles, preserving their shape and continuity without discretization. Additionally, BRT employs a topology-aware embedding method that organizes these geometric embeddings into a sequence of discrete tokens suitable for Transformers, capturing both geometric and topological characteristics within B-rep models. This enables the Transformer's attention mechanism to effectively learn shape patterns and contextual semantics of boundary elements in a B-rep model. Extensive experiments demonstrate that BRT achieves state-of-the-art performance in part classification and feature recognition tasks. Show Chinese abstract

Abstract: 近期，由 Transformer 网络驱动的生成式人工智能（AI）在自然语言处理、计算机视觉和图形学方面取得了显著的成功。然而，在计算机辅助设计（CAD）中应用 Transformer，特别是用于处理边界表示（B-rep）模型，仍然鲜有探索。为弥合这一差距，本文介绍了一种名为边界表示 Transformer（BRT）的新方法，该方法将 Transformer 适应于 B-rep 学习。由于 B-rep 模型具有不规则的拓扑结构和连续的几何定义，这些特性从根本上不同于 Transformer 设计时所针对的结构化和离散数据，因此带来了独特的挑战。为了解决这些问题，BRT 提出了一种连续几何嵌入方法，将 B-rep 表面（裁剪和未裁剪的）编码为 Bézier 三角形，从而在不进行离散化的情况下保留其形状和连续性。此外，BRT 还采用了一种拓扑感知嵌入方法，将这些几何嵌入组织成适合 Transformer 的离散标记序列，捕捉 B-rep 模型中的几何和拓扑特征。这使得 Transformer 的注意力机制能够有效地学习 B-rep 模型中边界元素的形状模式和上下文语义。大量实验表明，BRT 在零件分类和特征识别任务中达到了最先进的性能。