Action Recognition in Sports Training: An Exploratory Study Using Machine Learning Algorithms for X-Reality

With the gradual penetration of artificial intelligence technology into the field of education, improving learning efficiency and learner experience better has gradually become a research direction of great interest, especially in the sports training part of the physical education field. The addition of AI brings deep potential for the innovative development of XR. For youth basketball training, researchers conducted a series of model training through machine learning algorithms of artificial intelligence. The player's action behaviour and the trajectory of the ball were captured through image recognition to provide timely feedback to the player. The study results showed that the algorithm stopped training in the 69th to 81st round, indicating that the action recognition model set by the study reached a better balance here. It can maintain high accuracy in recognizing player behaviours and actions and provides a research basis for the subsequent implementation of continuous tracking and analysing behaviors of each player and their actions on the basketball court from start to finish. Afterwards, we discuss the future use of action recognition and extended reality to personalise player profiles in sports training, including the possibility of drawing training advice from multiple stakeholders to make player training more efficient and to increase the immersion of players in sports training.