A Semi-Supervised Learning-Based Method for Recognizing Volleyball Players' Arm Movement Trajectories.

In order to realize the recognition of athletes' arm trajectories with low data labeling cost, a semi-supervised learning-based method is proposed for volleyball players' arm trajectory recognition. A support vector machine framework is employed for the recognition of volleyball players' arm trajectories. To augment the dataset of volleyball sports samples and minimize the expense of data labeling, semi-supervised learning techniques are incorporated. The optimization of the support vector machine is combined with graph-based semi-supervised learning to develop a graph-based fuzzy least-squares support vector machine, and the classification results of graph-based fuzzy least-squares support vector machine are solved by the dyadic form and the representation theorem. Complete the training. Input the recognized volleyball player's movement data into the trained graph-based fuzzy least squares support vector machine, and output the recognition results of volleyball player's arm trajectory. The experimental results show that the method has the highest recognition accuracy when the width of the Laplace kernel function is 15, and the method can accurately lock the athlete's arm and track the athlete's arm movement in the recognition of a real game. [ABSTRACT FROM AUTHOR]