CSG Model

Constructive Solid Geometry (CSG) models represent 3D shapes by combining simpler primitives through Boolean operations (union, intersection, difference). Current research focuses on efficiently learning CSG representations from various data sources, including point clouds and meshes, using techniques ranging from deep learning and evolutionary algorithms to curriculum-based training of graph neural networks. These advancements aim to improve the accuracy and efficiency of 3D model generation and manipulation, with applications in areas such as computer-aided design (CAD), urban scene generation, and analysis of complex relationships represented by signed graphs. The development of more compact and robust CSG models is a key goal, leading to improved performance and reduced computational costs.