US 12,443,880 B2
Incremental training of a machine-learning model (MLM) for multiview item recognition
Christian Lee McDaniel, Atlanta, GA (US); Stefan Bjelcevic, Roswell, GA (US); Justin Paul, Atlanta, GA (US); Georgiy Pyantkovs'ky, Atlanta, GA (US); and Brent Vance Zucker, Roswell, GA (US)
Assigned to NCR Voyix Corporation, Atlanta, GA (US)
Filed by NCR Voyix Corporation, Atlanta, GA (US)
Filed on Apr. 29, 2022, as Appl. No. 17/733,139.
Application 17/733,139 is a continuation in part of application No. 17/665,145, filed on Feb. 4, 2022.
Prior Publication US 2023/0252343 A1, Aug. 10, 2023
Int. Cl. G06N 20/00 (2019.01); G06K 7/14 (2006.01)
CPC G06N 20/00 (2019.01) [G06K 7/1413 (2013.01)] 20 Claims
OG exemplary drawing
 
19. A system, comprising:
a server comprising at least one processor and a non-transitory computer-readable storage medium;
the non-transitory computer-readable storage medium comprises executable instructions; and
the executable instructions when executed by the at least one processor from the non-transitory computer-readable storage medium cause the at least one processor to perform operations comprising:
identifying portions of images captured of a transaction area for a transaction terminal during a checkout at the transaction terminal by an operator, wherein each image depicts multiple items within the transaction area from a different camera situated at a unique location within the transaction area and having a different perspective angle of the transaction area from remaining cameras associated with multiple images;
determining that a set of the portions that span the images is associated with an unknown item;
presenting the set of the portions on a display of the transaction terminal to the operator for the operator to identify a known item associated with the set of the portions from the multiple items;
instructing the operator to a scan of an item barcode for the known item identified by the operator at the transaction terminal;
receiving an item code associated with the known item after the scan;
receiving multiple additional images captured of the known item within the transaction area during the scan;
labeling the set of the portions within the images and the multiple additional images with the item code for the known item; and
saving the images with each portion of the set labeled with the item code along with the multiple additional images labeled with the item code as training images; and
training an item recognition Machine-Learning Model (MLM) on the training images during a training session.