US 11,900,558 B2
Reconstructing three-dimensional models of objects from real images based on depth information
Marissa Ramirez de Chanlatte, Los Altos, CA (US); Radomir Mech, Mountain View, CA (US); Matheus Gadelha, San Jose, CA (US); and Thibault Groueix, Grenoble (FR)
Assigned to Adobe Inc., San Jose, CA (US)
Filed by Adobe Inc., San Jose, CA (US)
Filed on Nov. 5, 2021, as Appl. No. 17/520,071.
Prior Publication US 2023/0147722 A1, May 11, 2023
Int. Cl. G06T 19/20 (2011.01); G06N 3/04 (2023.01); G06T 7/50 (2017.01); G06F 18/214 (2023.01); G06T 17/00 (2006.01)
CPC G06T 19/20 (2013.01) [G06F 18/214 (2023.01); G06N 3/04 (2013.01); G06T 7/50 (2017.01); G06T 17/00 (2013.01); G06T 2219/2004 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A non-transitory computer-readable medium storing instructions that, when executed by at least one processor, cause the at least one processor to:
determine a depth map for a real two-dimensional (2D) image depicting a digital object;
based on the depth map, generate a reconstructed three-dimensional (3D) model of the digital object comprising outer lines conforming to a shape of the digital object by utilizing a 3D-object-reconstruction-machine-learning model trained on a plurality of real 2D images to generate a predicted signed distance function based on the depth map and reconstructing the outer lines of the digital object based on the predicted signed distance function; and
provide, for display within a graphical user interface, a graphical depiction of the reconstructed 3D model or a visual effect derived from the reconstructed 3D model.