US 12,288,013 B2
Techniques for generating UV-net representations of 3D CAD objects for machine learning models
Pradeep Kumar Jayaraman, Toronto (CA); Thomas Ryan Davies, Toronto (CA); Joseph George Lambourne, London (GB); Nigel Jed Wesley Morris, Toronto (CA); Aditya Sanghi, Toronto (CA); and Hooman Shayani, London (GB)
Assigned to AUTODESK, INC., San Francisco, CA (US)
Filed by AUTODESK, INC., San Francisco, CA (US)
Filed on Jun. 15, 2021, as Appl. No. 17/348,295.
Claims priority of provisional application 63/169,070, filed on Mar. 31, 2021.
Prior Publication US 2022/0318466 A1, Oct. 6, 2022
Int. Cl. G06F 30/27 (2020.01); G06F 30/23 (2020.01); G06N 20/00 (2019.01); G06T 17/10 (2006.01)
CPC G06F 30/27 (2020.01) [G06F 30/23 (2020.01); G06N 20/00 (2019.01); G06T 17/10 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method for generating UV-net representations of three-dimensional (3D) computer-aided design (CAD) objects for machine learning models, the method comprising:
generating a first graph based on a first boundary-representation (B-rep) of a 3D CAD object;
discretizing a first parameter domain of a first parametric surface associated with the first B-rep into a first two-dimensional (2D) grid;
computing at least a first feature at a first grid point included in the first 2D grid based on the first parametric surface to generate a first 2D UV-grid; and
generating a first UV-net representation based on the first graph and the first 2D UV-grid.