This study aims to improve classification accuracy of different Support Vector Machine (SVM) models in classifying flat ground tricks namely Ollie, Kickflip, Shove-it, Nollie and Frontside 180 through the identification of significant time-domain features. An amateur skateboarder (23 years of age ±5.0 years` experience) executed five tricks for each type of trick repeatedly on a customized ORY skateboard (IMU sensor fused) on a cemented ground. From the IMU data a total of 36 features were extracted through statistical measures. The significant features were identified through two feature selection methods, namely Pearson and Chi-Squared. The variation of the SVM models (kernel-based) was evaluated both on all features and selected features in classifying the skateboarding tricks. It was shown from the study that all classifiers improved significantly in terms of training accuracy, prediction speed, training time and test accuracy. The Cubic-based SVM and Quadratic-based SVM demonstrated a 100% accuracy on both the test and train dataset, however, the Cubic-based SVM model provided the fastest training time and prediction speed between the two models. It could be concluded that the proposed method is able to improve the classification of the skateboarding tricks well.
© Copyright 2020 Embracing Industry 4.0. Lecture Notes in Electrical Engineering, 678. Veröffentlicht von Springer. Alle Rechte vorbehalten.