The shape variational autoencoder: A deep generative model of part‐segmented 3D objects

• Published in: Computer graphics forum Vol. 36; no. 5; pp. 1 - 12
• Main Authors: Nash, C., Williams, C. K. I.
• Format: Journal Article
• Language: English
• Published: Oxford Blackwell Publishing Ltd 01.08.2017
• Subjects: Categories and Subject Descriptors (according to ACM CCS); Finite element method; I.3.5 [Computer Graphics]: Computational Geometry and Object Modelling; Quantitative analysis; Reconstruction; Representations; Three dimensional models
• ISSN: 0167-7055; 1467-8659
• DOI: 10.1111/cgf.13240

We introduce a generative model of part‐segmented 3D objects: the shape variational auto‐encoder (ShapeVAE). The ShapeVAE describes a joint distribution over the existence of object parts, the locations of a dense set of surface points, and over surface normals associated with these points. Our mode...