FilletRec: A Lightweight Graph Neural Network with Intrinsic Features for Automated Fillet Recognition

TL;DR

This work tackles automatic fillet recognition and defeaturing in CAD for CAE by introducing FilletRec, a lightweight graph neural network that leverages intrinsic curvature ($K$) and mean curvature ($H$) together with topology and engineering attributes to robustly identify fillet faces. Built on a large-scale, diverse fillet-centric dataset, FilletRec achieves state-of-the-art accuracy with a fraction of the parameters of prior models, and demonstrates strong generalization across industrially complex models. The authors also integrate a robust Extend-Intersect-Clean defeaturing algorithm to convert detected fillets into sharp edges, enabling an end-to-end workflow from recognition to model simplification. Together, these contributions offer a practical, efficient pipeline that enhances CAE preprocessing and suggests directions for extending data-driven fillet and feature simplification in CAD pipelines.

Abstract

Automated recognition and simplification of fillet features in CAD models is critical for CAE analysis, yet it remains an open challenge. Traditional rule-based methods lack robustness, while existing deep learning models suffer from poor generalization and low accuracy on complex fillets due to their generic design and inadequate training data. To address these issues, this paper proposes an end-to-end, data-driven framework specifically for fillet features. We first construct and release a large-scale, diverse benchmark dataset for fillet recognition to address the inadequacy of existing data. Based on it, we propose FilletRec, a lightweight graph neural network. The core innovation of this network is its use of pose-invariant intrinsic geometric features, such as curvature, enabling it to learn more fundamental geometric patterns and thereby achieve high-precision recognition of complex geometric topologies. Experiments show that FilletRec surpasses state-of-the-art methods in both accuracy and generalization, while using only 0.2\%-5.4\% of the parameters of baseline models, demonstrating high model efficiency. Finally, the framework completes the automated workflow from recognition to simplification by integrating an effective geometric simplification algorithm.

Figures (16)

• Figure 1: Illustration of common edge features in a CAD model. The model on the left shows sharp concave and convex edges. The model on the right shows the same part after the application of fillets (on the concave edge), rounds (on the convex edges), and chamfers.
• Figure 2: Examples of fillet features.
• Figure 3: The overall pipeline of our proposed method. Given a B-Rep model, we first construct a graph by mapping faces to nodes and adjacency to edges. We then extract intrinsic geometric features (curvature) and attribute features (face width, dihedral angle) for each node. Finally, these features are fed into our FilletRec network for binary classification to identify fillet faces.
• Figure 4: Schematic diagram of face width $W_{\mathcal{S}}$ and dihedral angle $\theta_{\mathcal{S}}$ between adjacent faces.
• Figure 5: Classification of fillet features in the dataset. (a) Uniform-radius regular fillets. (b) Variable-radius regular fillets. (c) Irregular fillets.