МЕТОД ДЕТЕКТУВАННЯ ЛЮДЕЙ У НАТОВПАХ РІЗНОЇ ЩІЛЬНОСТІ

This work devoted for the analysis of the modern state-of-the-art solutions in crowd counting. Highlighted tendencies and widely used tools of this area. Special attention devoted to the crowd detection with the different density. Methods based on the Convolution Neural Networks analyzed and compared. Provided additional analyze for classic local binary pattern detection methods and brand-new state-of-the-art model YOLOv8, especially for low-dense crowd counting. Various methods compared based on their prospers and concerns.