USING OF NEURAL NETWORKS TO IMPROVE THE QUALITY OF SERVICE IN RETAIL CHAINS

This article is devoted to the development of a human posture classifier designed to monitor the actions of employees, such as cashiers, in order to improve customer service. Using a microservice based on the YOLO model, the system tracks customers and determines employee positions, recording situations of inactivity. The architecture of the classifier, based on the simplest perceptron, includes the processing of data on the "key points" of a person. The article also discusses the need to further train the model and methods to minimize false alarms in order to optimize the real-time operation of the system.