PNIPU Neural Networks to assess the quality of training for young football players

The main articles are:

• Decision Support Systems (DSS)
• Video analytics (terms, applications, technologies)

2025: Introducing a neural network to assess the quality of training for young football players

Scientists of the Perm Polytechnic proposed neural network technology to assess the quality of training for young football players. The university announced this on January 27, 2025.

With the development of methods computer vision , opportunities appeared for analyzing and improving the quality of training for young people. athletes One of the tools in this area is the technology of three-dimensional detection - the determination in space and time of the position of key points of a person. It allows not only to track movements, but also to conduct a deep analysis of their technique, helping coaches and athletes to identify weaknesses, improve skills and automate quality control of exercise. Scientists of the Perm Polytechnic University have developed a prototype of an information system for supporting coaching solutions based on neural network technology. This will allow evaluating the training process of football players using intellectual analysis data obtained from video cameras.

You can determine the movements of an athlete by the position of his 2D skeleton and interaction with sports equipment. But this is not always enough, since for some exercises you need to know the location of key points relative to each other in space. Moreover, after filming, you will need to synchronize personnel from the left and right cameras in time. When using neural networks in 3D space, this is no longer necessary. In general, the 3D approach offers better accuracy and depth of analysis compared to two-dimensional detection methods, so it is the basis for a computer program.

Scientists of the Perm Polytechnic with the help of a trained neural network have developed a prototype of a decision support information system that can determine how well the training of young football players is going on. The system allows you to track the individual work of each athlete of the team at the same time and automate quality control on the part of the coach.

In total, the program fixes 34 key points of a person, including shoulders, elbows, hands, fingers and toes, hip joints, knees and feet. Video cameras are installed on the training field, and the software and hardware system records exercises in the form of a video sequence and transmits it to a computer, where errors are detected when performing exercises with and without the ball. This will allow coaches and analysts to conduct a detailed analysis of the technique of members of the football team and develop strategies for improving sportsmanship, "said Alexander Terekhin, graduate student of the Department of Computational Mathematics, Mechanics and Biomechanics at PNIPU.

To check the operation of the system, we conducted experiments on a number of exercises that require the analysis of three-dimensional images, for example, slopes. The task of the neural network is to determine how much the quality of the athlete's movements meets the specified requirements: do not bend his legs in his knees, touch the floor with the fingers of both hands for no more than 3 seconds, etc. Video shooting of the player was carried out on the right so that there was no overlap of some parts of the body with others, which is why the neural network may not understand how to combine separate key points into a skeleton. According to preliminary results, the developed technology fully copes with the detection of errors in the player's movements, - explained Valery Stolbov, head of the Department of Computational Mathematics, Mechanics and Biomechanics, PNIPU, Doctor of Technical Sciences.

In the future, it is planned to expand the number of sports exercises analyzed (at least 40) and conduct comprehensive tests in the football arena in the process of real training. The development of scientists from the Perm Polytechnic will increase the efficiency of classes and automate the process of processing the results of testing young football players by introducing computer vision and artificial intelligence methods.