CADLabel: Development of an annotation tool for deep learning tasks on 3D CAD models

The recognition of features is important in computer-aided engineering (CAE) across different applications, such as computer aided process planning (CAPP) or computer aided manufacturing (CAM). While deep learning has become a widely used approach for the recognition of features, it necessitates large datasets that are still difficult to compile. To facilitate the annotation of 3D CAD models, a novel annotation tool is introduced. The CADLabel annotation tool offers direct and indirect manual labeling modules for 3D CAD models, but it also contains a recommendation system leveraging geometrical and topological analysis with a graph neural network (GNN) to propose potential features, streamlining the annotation workflow. Additional attributes include support for multiple STEP file formats, integration to add further information to the 3D CAD model such as Product and Manufacturing Information (PMI), and the ability to export annotations in different file formats suitable for deep learning.