A Machine Learning Based Method to Establish Normative Action Model for Table Tennis.

In recent years, how to apply machine learning methods to analyze and evaluate the movement of athletes or amateurs in sports videos becomes a hot research topic. In this paper, we design a machine learning based method to establish a normative action model of table tennis by analyzing collected training videos. In addition, we build a normative action data set from training videos of professional athletes and a comparative experimental data set for from training videos of amateurs. We define 10 action judgment rules according to suggestions from professional coaches and athletes, and verify the effectiveness of the proposed normative action model by experimental evaluations. We hope that the proposed method and built data sets can inspire more excellent work for action analysis of sports.