Motion Classification by Utilizing Machine Learning for Acceleration Data
Kenzaburo Ishizu, Toshimitsu Honda, Kohei Miyase, Stefan Holst, Shudai Ishikawa, Roshwin Sengupta, Ilia Polian · 2025
In recent years, professional athletes have utilized motion analysis to improve their performance and skill. However, since devices and systems used in motion analysis are very expensive, it is impossible for an athlete who does not have enough budget to use them. In this paper, we propose a motion analysis system using an inexpensive accelerometer controlled by Arduino and machine learning. The target motion in this work is bat swing in baseball. The acceleration from the target motion is measured and used to train a neural network. Users of this system can classify their motion into experienced one and inexperienced one, and then will be able to know the quality of their swing motion.