CAD segmentation datasets are predominantly synthetic or automatically labelled, leaving it unclear how well existing methods transfer to real-world engineering tasks. Fastener-relevant hole detection represents one such task, requiring human annotation and exhibiting greater geometric complexity than commonly used benchmarks. We address this gap through three main contributions. First, we develop two task-specific rule-based methods, one graph-wire-search and one multi-view based, and complement them with a human-annotated benchmark dataset derived from engineering CAD data that also enables training of UV-Net and AAGNet, representative learning-based approaches for B-Rep segmentation. Second, we conduct an extensive experimental study including parameter optimisation, evaluation on our benchmark dataset, cross-domain generalisation on a synthetic dataset, and an assessment of the feasibility of the approaches at varying dataset sizes and IoU thresholds. Third, we provide evidence-based insights into the strengths and limitations of different approaches: rule-based methods generalise well without training data, with the graph-wire-search method favouring recall and the multi-view method favouring precision, whereas learning-based methods achieve strong performance with sufficient annotations but exhibit reduced robustness under domain shift. These findings highlight trade-offs between data requirements, and robustness, and provide guidance not only for hole detection, but more broadly for real-world CAD segmentation tasks.
Figures & tables
Figure 1: Overview of screws and fastener holes
Figure 2: Comparison of publicly available CAD segmentation datasets. Top: face-type distribution. Bottom: average number of faces per part.
Per-hole
Per-face
Method
Recall
Prec.
F1
Recall
Prec.
F1
AAGNet
0.944
0.810
0.871
0.912
0.932
0.922
UV-Net
0.934
0.896
0.914
0.883
0.933
0.920
GWS
0.927
0.578
0.712
0.977
0.644
0.733
MVR
0.728
0.870
0.793
0.768
0.898
0.871
Table 1: Results with models trained and tested on our dataset, reported per-hole and per-face
Per-hole
Per-face
Method
Recall
Prec.
F1
Recall
Prec.
F1
AAGNet
0.013
0.357
0.026
0.023
0.433
0.044
UV-Net
0.468
0.933
0.623
0.381
0.957
0.545
GWS
0.889
0.805
0.845
0.983
0.768
0.871
MVR
0.617
0.987
0.760
0.577
0.987
0.729
Table 2: Generalisation study with models trained on our dataset and tested on MFCAD++.
Figure 3: Recall and precision values for all methods at different training dataset sizes
Figure 4: Recall and precision at different IoU thresholds on HDD (top) and MFCAD++ (bottom).
Figure 5: Parameter optimisation results for the graph-wire-search method
Figure 6: Parameter optimisation results of the Multiview Rendering method
Figure 7: Recall and precision for AAGNet (top) and UV-Net (bottom) under different class weights and oversampling ratios.
Appendix figures & tables2 assets
Supplementary material from the paper’s appendix.
Appendix
Statistic
Count / Ratio
Hole faces
14,471
Models
2,202
Total faces
197,003
Hole face proportion
0.073
Holes
6,864
Avg. holes per model
3.02
Appendix
Table 3: Dataset statistics
Figure 8: Representative failure modes. Green indicates correctly detected holes, red indicates missed holes, and yellow indicates detected but unlabelled holes.
Deep learning in computer-aided design (CAD) remains fundamentally constrained by the data scarcity challenge: authentic CAD data is difficult to collect at scale, while synthetic data may not faithfully reflect real design practice. Rather than pursuing ever-larger CAD datasets, this paper alternatively treats CAD learning as a knowledge completion and calibration problem. It introduces KDH-CAD, a knowledge-data hybrid framework that integrates pretrained knowledge in foundation models, structured domain knowledge from textbooks/tutorials, and a very small amount of labeled CAD data. Domain knowledge is used to elicit and complete CAD-relevant concepts that are weakly expressed or under-represented in pretrained foundation models, while labeled CAD data calibrates these concepts in the latent space to account for task-specific geometric variability, without fine-tuning the foundation model. Experiments on real-world mechanical part classification show that KDH-CAD achieves strong performance in low-data regimes, reaching 92.6% accuracy with only 250 training samples, 95.8% with 1,000 samples, and continuing to improve with additional data. This matches or exceeds state-of-the-art performance that typically requires an order of magnitude more data. These results suggest that combining pretrained foundation models with structured domain knowledge can substantially reduce reliance on large-scale CAD datasets, providing a principled and practical direction for data-efficient CAD learning.
Ziqin Gao, Zhijie Yang, Qiang Zou
State Key Laboratory of CAD&CG, Zhejiang University, Hangzhou, 310027, China
Learning directly from boundary representations (B-reps) has significantly advanced 3D CAD analysis. However, state-of-the-art B-rep learning methods rely on absolute coordinates and normals to encode global context, making them highly sensitive to rotations. Our experiments reveal that models achieving over 95% accuracy on aligned benchmarks can collapse to as low as 10% under arbitrary SO(3) rotations. To address this, we introduce FoV-Net, the first B-rep learning framework that captures both local surface geometry and global structural context in a rotation-invariant manner. Each face is represented by a Local Reference Frame (LRF) UV-grid that encodes its local surface geometry, and by Field-of-View (FoV) grids that capture the surrounding 3D context by casting rays and recording intersections with neighboring faces. Lightweight CNNs extract per-face features, which are propagated over the B-rep graph using a graph attention network. FoV-Net achieves state-of-the-art performance on B-rep classification and segmentation benchmarks, demonstrating robustness to arbitrary rotations while also requiring less training data to achieve strong results.
Verbal-prompted segmentation is inherently limited by the expressiveness of natural language and struggles with uncommon, instance-specific, or difficult-to-describe objects: scenarios frequently encountered in manufacturing and 3D printing environments. While image exemplars provide an alternative, they primarily encode appearance cues such as color and texture, which are often unrelated to a part's geometric identity. In industrial settings, a single component may be produced in different materials, finishes, or colors, making appearance-based prompting unreliable. In contrast, such objects are typically defined by precise CAD models that capture their canonical geometry. We propose a CAD-prompted segmentation framework built on SAM3 that uses canonical multi-view renderings of a CAD model as prompt input. The rendered views provide geometry-based conditioning independent of surface appearance. The model is trained using synthetic data generated from mesh renderings in simulation under diverse viewpoints and scene contexts. Our approach enables single-stage, CAD-prompted mask prediction, extending promptable segmentation to objects that cannot be robustly described by language or appearance alone.
Zhenran Tang, Rohan Nagabhirava, Changliu Liu
the Robotics Institute, Carnegie Mellon University, Pittsburgh, PA, USA