Abstract
This paper examines the development of continuous signal pre-processing methods for segmentation, with a particular emphasis on the application of machine learning. Signal segmentation plays a key role in various fields such as medical diagnostics, image and video processing, and security and surveillance systems. The main focus is on improving the quality of segmentation through data pre-processing, which improves the accuracy and reliability of the final results. As an example, an image segmentation algorithm is presented using the U-Net architecture and the VOC 2012 dataset. The paper also presents a literature review covering current research and development in the field of signal pre-processing and signal segmentation.