CPC G06Q 20/208 (2013.01) [G06K 7/10722 (2013.01); G06K 7/1413 (2013.01); G06V 10/74 (2022.01)] | 8 Claims |
1. A method comprising:
scanning a first item using a first barcode scanner and a second barcode scanner, wherein the first barcode scanner and the second barcode scanner are arranged such that beams of the first barcode scanner and beams of the second barcode scanner are in different planes;
determining, based on signals from the first barcode scanner and the second barcode scanner, a first two-dimensional light grid indicating where the first item broke the beams of the first barcode scanner while breaking the beams of the second barcode scanner;
predicting, using a machine learning model and based on the first two-dimensional light grid, a first category for the first item, wherein the first two-dimensional light grid indicates to the machine learning model a size and shape of the first item, and wherein the first category of the first item is one of produce or a boxed item;
receiving, from a camera, an image of the first item;
after predicting the first category for the first item, retrieving, from a database, a first set of images of items assigned to the first category;
after retrieving the first set of images assigned to the first category, comparing the image from the camera of the first item to the first set of images assigned to the first category;
selecting an image from the first set of images of items in the first category based on similarities between (i) the image from the first set of images of items assigned to the first category and (ii) the image of the first item, wherein the image from the first set of images of items assigned to the first category shows an item;
determining an identity of the first item as the item in the image from the first set of images;
determining, based on a motion of a hand, that a second item is being taken by the hand; and
generating a message indicating that the second item is being taken.
|