US 12,272,044 B2
Production line conformance measurement techniques using categorical validation machine learning models
Thomas R. Gilbertson, Hartford, CT (US); Raja Mukherji, Dublin (IE); Haylea Tricia Northcott, Simsbury, CT (US); Karen Harte, Avon, CT (US); and Colby A. Wright, Chandler, AZ (US)
Assigned to Optum, Inc., Minnetonka, MN (US)
Filed by Optum, Inc., Minnetonka, MN (US)
Filed on Sep. 15, 2021, as Appl. No. 17/475,987.
Claims priority of provisional application 63/171,779, filed on Apr. 7, 2021.
Prior Publication US 2022/0327689 A1, Oct. 13, 2022
Int. Cl. G06T 7/00 (2017.01); G06F 18/21 (2023.01); G06N 3/08 (2023.01); G06T 7/11 (2017.01)
CPC G06T 7/0008 (2013.01) [G06F 18/217 (2023.01); G06N 3/08 (2013.01); G06T 7/11 (2017.01); G06T 2207/20132 (2013.01)] 16 Claims
OG exemplary drawing
 
1. A computer-implemented method comprising:
determining, by one or more processors, a plurality of image attributes and a validation category for an input production line image, wherein: (i) the plurality of image attributes comprises a product type identifier for the input production line image and a physical production attribute for the input production line image, and (ii) the validation category is characterized by a defined subset of the plurality of image attributes;
determining, by the one or more processors, based at least in part on the physical production attribute, and by utilizing a categorical validation machine learning model that is associated with the validation category, a validation prediction for the input production line image, wherein (i) the categorical validation machine learning model is generated using a transformed training production line image associated with the validation category, and (ii) the transformed training production line image is generated by (a) determining an image transformation for a training production line image based at least in part on the validation category and (b) applying the image transformation to the training production line image to generate the transformed training production line image; and
initiating, by the one or more processors, a prediction-based action based at least in part on the validation prediction.