МЕТОДИ СЕМАНТИЧНОЇ СЕГМЕНТАЦІЇ ЗОБРАЖЕННЯ ТА ЇХ ПОРІВНЯННЯ

This work is devoted to research, comparison and implementation of semantic image segmentation methods with the aim of developing an effective algorithm for automatic selection and classification of objects in images. The purpose of the study is to determine the optimal method or combination of methods that provide the highest segmentation accuracy and speed of image processing. Convolutional Neural Networks (CNNs) were chosen as the main method, noting their high accuracy, generalizability and efficiency in different settings. The implementation of the selected method is performed using a framework for deep learning. The research results can be useful for various applications in modern image processing systems and intelligent systems.