Tennis stroke detection using inertial data of a smartwatch

To assist individuals in sports activities is one of the emerging areas of wearable applications. Among various kinds of sports, detecting tennis strokes faces unique challenges. In this sport the speed of strokes is high, enforcing wearable sensors to have high sampling rates, high-speed bus (to transfer the data to the processor), and the most importantly adequate size of high-speed memory. The constraints encourage researchers to design a custom made hardware to cope with the challenges. The research question that we are trying to address is to show how accurate a commercial smartwatch can detect tennis strokes by using various techniques in machine learning. In this paper, we propose an approach to detect three tennis strokes by utilizing a smartwatch. In our method, the smartwatch is part of a wireless network in which inertial data file is transferred to a laptop where data prepossessing and classification is performed. The data file contains acceleration and angular velocity data of the 3D accelerometer and gyroscope. We also enhanced our method with data prepossessing techniques to elevate data quality. The evaluation of our devised method shows promising results compared to a similar method.