Scenario characterization using machine learning user tracking and profiling for a cashier-less retail store

摘要

Devices, systems, and method are provided for tracking items in a store for processing a cashier-less purchase transaction. In one example, a method includes tracking a shopper in a store using one or more sensors to process an account of shopping activity performed by the shopper. The method further includes monitoring the shopper in the store using said one or more sensors. Output of the one or more sensors is processed by a processing entity associated with the store to produce input feature data for aspects of a current scenario involving actions of the shopper and one or more items in the store. The processing entity associated with the store processes the input feature data to characterize aspects of the current scenario. The processing entity characterizes aspects of the current scenario in part by accessing a trained machine learning model that is maintained for a profile of the shopper. The input feature data is for actions of the shopper in relation to the scenario. The input feature data is processed by one or more classifiers of the trained machine learning model to produce the characterized aspects of the scenario relating to the account of said shopping activity of the shopper and said one or more items in the store.