On the Issue of Comparing the Effectiveness of YOLOv8 and U-Net Neural Networks in the Tasks of Segmentation of Territorial Objects

There are many problems associated with the automation of aerial photography. Solutions based on machine learning algorithms turned out to be particularly effective for solving them. However, choosing a suitable computer vision model is a difficult task due to the variety of models available. The purpose of this scientific work is to select the most popular neural network architectures and identify the most effective architecture in terms of efficiency and quality of work performed.