US 11,055,573 B2
Generating synthetic models or virtual objects for training a deep learning network
Reza Farivar, Champaign, IL (US); Kenneth Taylor, Champaign, IL (US); Austin Walters, Savoy, IL (US); Joseph Ford, III, Manakin-Sabot, VA (US); and Rittika Adhikari, Westord, MA (US)
Assigned to Capital One Services, LLC, McLean, VA (US)
Filed by Capital One Services, LLC, McLean, VA (US)
Filed on Aug. 15, 2019, as Appl. No. 16/541,406.
Application 16/541,406 is a continuation of application No. 16/250,719, filed on Jan. 17, 2019, granted, now 10,430,692.
Prior Publication US 2020/0234083 A1, Jul. 23, 2020
Int. Cl. G06K 9/62 (2006.01); G06N 3/08 (2006.01)
CPC G06K 9/6262 (2013.01) [G06K 9/6256 (2013.01); G06N 3/08 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
identifying, by one or more devices, data for generating a set of synthetic models of an object,
wherein the data identifies:
positions that a plurality of components, of a synthetic model in the set of synthetic models, are permitted to assume, and
a permissible size of a component of the plurality of components,
wherein the plurality of components include:
a first component corresponding to a first part of the object, and
a second component corresponding to a second part of the object, and
wherein the component of the plurality of components is the first component or the second component;
generating, by the one or more devices, the set of synthetic models based on the data,
wherein generating the set of synthetic models includes:
generating the synthetic model such that no portions of the first component and the second component overlap one another; and
causing, by the one or more devices, the set of synthetic models to be provided to a deep learning network to train the deep learning network to perform at least one of image segmentation, object recognition, or motion recognition.