Machine Learning based Movement Analysis and Correction for Table Tennis1

Abstract

Table tennis is a popular sport with high popularity in the world. Owing to the limited number of professional coaches, most of table tennis amateurs expect to have movement guidance by artificial intelligence. However, existing researches on table tennis movements mainly focus on the classification of strokes, which can hardly help amateurs correct their wrong movements. To solve this problem, we propose a quantitative analysis and correction method of table tennis movement based on machine learning. In this method, we design a set of evaluation metrics to quantify players’ movements and provide correction suggestions to them. In addition, we built a dataset of table tennis movement analysis and correction. Based on this dataset, we verify the effectiveness of the proposed method with high-performance indicators. We hope our work and the dataset can inspire more excellent research works on quantitative analysis and correction of movements in table tennis.