Recognition of the Basketball Players Position Using Live Cameras

This article discusses the research aimed at creating a program that will track the players of the attacking basketball team in various matches and transfer them to a minimap in the form of dots, showing the trajectory of each player's movement. The article also reviews previous research on this topic, which partially solve the tasks set, and describes the technologies used in this study. Models of previous works have been changed for this research. An experiment was presented that shows the trainability of the neural network that determines the players on the field. The article also describes a ball tracking experiment based on the trainability of neural networks. A study on transferring tracked data (players, ball) to a 2D diagram of a basketball field was conducted.