FootbSense: Soccer Moves Identification Using a Single IMU

Abstract

Although wearable technologies are commonly used for sports at elite levels, these systems are expensive, and it is still difficult to recognize detailed player movements. We introduce a soccer movements recognition system using a single wearable sensor to aid the skill improvement for amateur players. We collected 3-axis acceleration data of six soccer movements and validated the proposing system. We also compared three sensor locations to find the best accurate location. With ensemble bagged trees classification method, we achieved 78.7% classification accuracy of six basic soccer movements from the inside-ankle sensor. Moreover, our results show that it is possible to distinguish between running and dribbling, passing and shooting, even though they are similar movements in soccer. Besides, the second highest accuracy was achieved from a sensor placed on the upper part of the back, which is a safer wearing position compared to other locations. These results suggest that our approach enables a new category of wearable recognition system for amateur soccer.