Аннотация
The implementation of robotic systems and digitalization in agriculture are important tasks today. In this paper, the possibility of using pattern recognition and machine learning methods as a computer model of an agricultural robot for harvesting is considered. The grading of tomato fruits can be classified based on their ripeness according to their life cycles, which can be identified by their color: green in the growing stage, yellow in the pre-ripening stage, and red when ripe. Conventional skill-based methods cannot meet the exact selection criteria for modern production management in the agricultural sector as they are time-consuming and of low accuracy. Automatic feature extraction behavior using machine learning is most effective in image classification and recognition tasks. Thus, the article presents the results of a study on the recognition of ripe tomato fruits by a robotic system, carried out within the framework of the grant project of the Ministry of Education and Science of the Republic of Kazakhstan AP08857573 and implemented classical algorithms based on the HSV color model and color segmentation using the k-means algorithm as comparative algorithms and based on machine learning, a universal intelligent tomato classification system is proposed for practical use using Yolo 5. This study aims to provide an inexpensive solution with the best performance and accuracy for assessing tomato ripeness. The results are collected in terms of accuracy, loss curves and confusion matrix. The results showed that the proposed model outperforms other machine learning (ML) methods used by researchers for tomato classification problems, providing 99% accuracy.