The use of neural networks and ballistic models to enhance the effectiveness of ball-throwing training

the article analyzes the possibilities of applying neural networks and modern technical solutions to optimize and accelerate ball throwing training. This skill helps to develop student's strength, endurance, accuracy, coordination and reaction speed. A simple system for optimizing the process of training students in small ball throwing skill is proposed, which consists of three blocks. The fixation block allows representing the position of the student's body parts in the form of a graph based on the analysis of photo or video of the throwing moment with the use of neural network YOLOv8-Pose. The second block of the system performs theoretical analysis and verification of the throwing distance. This block is based on a simple ballistic physical and mathematical model, and verification is based on real throws of the ball by the student. The third block generates personalized instructions for adjusting the body position to increase the throwing distance using the k-closest method. The pedagogical experiment on the use of the simple system in the classroom showed: in the experimental group there was a significant increase in the speed of ball flight and acceleration dynamics for the final effort compared to the control group. The validity of the results was confirmed by statistical processing using Student's t-criterion. The use of even a simple version of this system in the classroom allows the student and the teacher to conduct an objective analysis of technical errors in throwing a small ball visually, and the lesson itself becomes exciting for the student and allows to work out in more detail the errors made by the student when throwing. It is shown that the integration of digital tools even in a simple version in the educational process makes the learning process more interactive and focused on personal development.

neural network technologies, video analytics, ball throwing, teaching