ExtrudeNet: Unsupervised Inverse Sketch-and-Extrude for Shape Parsing

Sketch-and-extrude is a common and intuitive modeling process in computer aided design. This paper studies the problem of learning shape given form point clouds by “inverse” sketch-and-extrude. We present ExtrudeNet, an unsupervised end-to-end network for discovering sketch extrude from clouds. Behind ExtrudeNet are two new technical components: 1) effective representation extrude, which can model extrusion with freeform sketches conventional cylinder box primitives as well; 2) numerical method computing signed distance field used learning. first attempt that uses machine to reverse engineer sketch-and-extrude fashion. not only outputs compact, editable interpretable be seamlessly integrated into modern CAD software, but also aligns standard facilitating various editing applications, distinguishes our work existing parsing research. Code released at https://github.com/kimren227/ExtrudeNet .