REVIEW OF EFFECTIVE IMAGE SEGMENTATION USING DATA CLUSTERING METHODS

Abstract

Relevance. For many decades, people have dreamt of creating machines that could match human intelligence, capable of thinking and acting like humans. One of the most fascinating ideas was to enable computers to "see" and interpret the surrounding world in a way that is comprehensible to them. Thanks to progress in the field of artificial intelligence and innovations in deep learning and neural networks, this area has made a significant leap in recent years, surpassing humans in some tasks related to object detection and labeling. The purpose of this work is to review and comparatively analyze existing image segmentation tools, particularly those using data clustering methods. The object of research is image segmentation algorithms, which are key in the field of computer vision for object recognition and analysis. The subject of research is the specific use of clustering methods in these algorithms, their effectiveness, and accuracy. Results. This paper conducts a review and comparative analysis of the main methods of data clustering. It highlights 3 of the most common and promising algorithms for improvement. For each of them, the working principle is briefly described, and their main advantages and disadvantages are highlighted. Conclusion. Among the methods considered, there is no best universal choice, as the effectiveness of each depends on the specific task. The described algorithms are planned to be used for their further research and modifications.