IMPROVEMENT OF THE SORTING PROCESS USING A COMPUTER VISION SYSTEM AND A LIGHT SENSOR IN SORTING AGRICULTURAL PRODUCTS

This article considers the issue of improving the sorting process of agricultural products, in particular fruits and vegetables. Due to the low accuracy and subjectivity of traditional sorting methods, the need to use modern technologies is justified. The study proposes an integrated approach to determine the external characteristics of products (color, shape, size) using a computer vision system and to evaluate internal quality indicators (ripeness, chemical composition) using near-infrared (NIR) spectroscopy. Also, a statistical selection method and a prediction model based on PLS and SVM algorithms were developed for evaluating product groups. The proposed system allows to improve the quality of export-oriented products, automate the sorting process and reduce the human factor. As a result, the sorting accuracy increases, the export of low-quality products decreases, and the overall production efficiency increases.