Дослідження методів обробки зображень. Методи визначення об’єктів

Abstract

This study examines image processing and object detection methods, highlighting the transition from traditional techniques to advanced machine learning and deep learning models like CNNs. Despite CNNs’ flexibility in complex scene analysis, their high data and computational demands present challenges. The Segment Anything Model (SAM) addresses these limitations by excelling in diverse conditions and continuously improving with new data. This work focuses on innovative automatic segmentation algorithms and simplifying complex image analysis by using point coordinates for input to produce segmented objects