Badminton Smashing Recognition through Video Performance by using Deep Learning

Authors

  • Zi Ying Yip Faculty of Manufacturing and Mechatronics Engineering Technology, Universiti Malaysia Pahang, 26600 Pahang, Malaysia.
  • Ismail Mohd Khairuddin Faculty of Manufacturing and Mechatronics Engineering Technology, Universiti Malaysia Pahang, 26600 Pahang, Malaysia.
  • Wan Hasbullah Mohd Isa Faculty of Manufacturing and Mechatronics Engineering Technology, Universiti Malaysia Pahang, 26600 Pahang, Malaysia.
  • Anwar P. P. Abdul Majeed Faculty of Manufacturing and Mechatronics Engineering Technology, Universiti Malaysia Pahang, 26600 Pahang, Malaysia.
  • Muhammad Amirul Abdullah Faculty of Manufacturing and Mechatronics Engineering Technology, Universiti Malaysia Pahang, 26600 Pahang, Malaysia.
  • Mohd Azraai Mohd Razman Faculty of Manufacturing and Mechatronics Engineering Technology, Universiti Malaysia Pahang, 26600 Pahang, Malaysia.

DOI:

https://doi.org/10.15282/mekatronika.v4i1.8607

Keywords:

Badminton, Smashing, Deep Learning, Accuracy, Confusion matrix

Abstract

Nowadays, badminton become the hot trends sport in Malaysia due to the influence of Lee Zii Jia which is the Malaysian badminton player and he has been participate the men’s single badminton in Tokyo 2020 Olympic Game at the Musashino Forest Sports Plaza in Tokyo. Due to this reason, sport analysis become one major contribution in analysing and improving the performance of athlete. Hence, this project constructs a badminton smashing recognition through video performance by using the deep learning. The main purpose of this project is to evaluate the performance of the models in classifying the types of smashing in badminton. The models will be trained using Deep Learning models of ResNet-18, GoogleNet and VGG-16 and the best precision of badminton smashing accuracy were compared. In this project, we found that ResNet-18 has the best performance of accuracy of 97.51% and 98.86% on both training and testing datasets respectively by using the software Jupyter. On other hand, GoogleNet has the highest accuracy of 83.04% and 97.20% on both training and testing datasets respectively by using hardware Jetson Nano.

Downloads

Published

2022-06-24

How to Cite

[1]
Z. Y. Yip, I. Mohd Khairuddin, W. H. Mohd Isa, A. P. P. Abdul Majeed, M. A. . Abdullah, and M. A. Mohd Razman, “Badminton Smashing Recognition through Video Performance by using Deep Learning”, MEKATRONIKA, vol. 4, no. 1, pp. 70–79, Jun. 2022.

Issue

Section

Original Article

Most read articles by the same author(s)

1 2 3 > >>