Parametric point clouds are sampled from CAD shapes, have become increasingly prevalent in industrial manufacturing. However, most existing point cloud learning methods focus on the geometric features, such as local and global features or developing efficient convolution operations, overlooking the important attribute of constraints inherent in CAD shapes, which limits these methods' ability to fully comprehend CAD shapes. To address this issue, we analyzed the effect of constraints, and proposed its deep learning-friendly representation, after that, the Constraint Feature Learning Network (CstNet) is developed to extract and leverage constraints. Our CstNet includes two stages. The Stage 1 extracts constraints from B-Rep data or point cloud. The Stage 2 leverages coordinates and constraints to enhance the comprehend of CAD shapes. Additionally, we built up the Parametric 20,000 Multi-modal Dataset for the scarcity of labeled B-Rep datasets. Experiments demonstrate that our CstNet achieved state-of-the-art performance on both public and proposed CAD shapes datasets. To the best of our knowledge, CstNet is the first constraint-based learning method tailored for CAD shapes analysis.

本研究针对现有点云学习方法忽视CAD形状固有约束的问题，提出了基于约束的深度学习模型CstNet，旨在更好地理解CAD形状。CstNet通过提取B-Rep数据或点云中的约束，并结合坐标信息，显著提升了对CAD形状的理解。实验结果表明，CstNet在公共和新提出的CAD形状数据集上均达到了领先的性能。

参数化点云的约束学习