Feasibility Study of Using Hybrid Artificial Intelligence Architecture for Recognizing Execution Styles of Players in Tennis Match

Abstract

A feasibility study of using a hybrid artificial intelligence architecture is presented for identifying basic tennis stroke types and players’ positions with tennis court labels by collecting the statistical data for studying player execution styles in a tennis match. The hybrid architecture consists of a machine-learning-based system for video processing and a rule-based system for identifying tennis strokes and players’ positions. It utilizes less computing resources than entirely machine-learning-based approaches. This architecture performs spatiotemporal information extraction to understand the players’ style, such as the time and court positions of classified tennis strokes.