This article presents the development of an automated system for monitoring the quality of dragees using machine vision and digital simulation technologies. The goal of this work is to create an intelligent system that can detect surface defects of confectionery products, such as chips and cracks, in real time. The proposed approach combines the training of a neural network based on the YOLO architecture using both real and synthetic images, as well as the implementation of digital simulation of the production line for pre-debugging and optimizing the system parameters. The use of a digital model allows testing in conditions close to real, which contributes to an increase in the accuracy of defect classification and a decrease in the costs of equipment setup. The tests carried out confirmed the high efficiency of the proposed system and the expediency of its implementation in the food industry.

Keywords: machine vision, quality control, dragees, YOLO, digital simulation, surface defects