Software and Hardware Complex Development toMonitor Production Activities Based on the YOLOv8 Neural Network

Abstract

The article discusses the features of software and hardware complex development to monitor the efficiency of employees by means of an ESP8266 microcontroller and video cameras with a YOLOv8 neural network that meets the requirements of a 'smart shop'. The purpose of this study is to justify and evaluate the working concept of the created software and hardware complex in the form of a working prototype. Examples of current developments of similar monitoring systems realized within the paradigm of the Internet of Things, are given. However, the systems found in international practice focus on processing images of moving personnel only and do not analyze the technical parameters of the equipment this personnel work on. This study attempts to overcome this limitation. A schematic diagram of the system operation consisting of four key modules has been developed: a data collection module from machine-tools, an employee identification module, a primary data processing module, a web server with software application to manage the entire system in real time. The MQTT (Message Queuing Telemetry Transport) protocol is used for communication between devices and the server in the designed system to provide quality of data transmission under conditions of channel bandwidth limitations. Combined analysis of data from both machine-tools (control of the load of the electric motor) and the movement of workers in the space of the production room is proposed within the concept of developing such systems for the first time. Evaluation tests on the monitored data showed the operability of the entire complex, the problem of object automatic detection in the camera image flow and their location within the coordinate system of the production room was solved.