Badminton Action Recognition Model Combining Artificial Neural Network Algorithm and DTW Algorithm

Authors

  • Yuanji Zhong Department of Social Sports Guidance and Management, School of Physical Education and Art, Jiangxi University of Science and Technology, Ganzhou, 341000, China
  • Jiangwei Yang Department of Physical Education, Xiamen Institute of Technology, Xiamen, 361021, China
  • Wenhao Guo Department of Social Sports Guidance and Management, School of Physical Education and Art, Jiangxi University of Science and Technology, Ganzhou, 341000, China

Abstract

Badminton is a sport that is susceptible to the influence of the racket and the player’s attire, and it is difficult to recognize the action without the help of 3D-capturing technology. To accurately recognize the technical movements involved in badminton, this study proposes a badminton action recognition model that combines an artificial neural network algorithm and a dynamic time warping algorithm. The badminton player’s posture can be estimated using the improved OpenPose algorithm, which can accurately capture the player’s key skeletal nodes. Based on this, the mobile network V3 network architecture and dynamic time warping algorithm are used to recognize the action. The results revealed that the mobile network V3 model achieved significant performance in recognizing badminton movements, with an accuracy of 0.987, which was significantly higher than the results obtained with the visual geometric group network, the residual network and the mobile network V2 model. Moreover, the total number of parameters was significantly decreased by 4.7M. In addition, the precision of the model was improved by an average of 12.27%, outperforming the other three network models. The dynamic time warping algorithm also performed well in evaluating badminton technical movements. The results of the evaluation were significantly improved by introducing the weighting values. In the smash and split score, the score based on dynamic time warping algorithm was 84, while the score after introducing the weights reached 86, which was closer to the scoring value of the domain experts. The results demonstrated that the mobile network V3 model with the dynamic time warping algorithm used in the study was able to achieve accurate recognition
of badminton movements with high computational efficiency. The study provides an efficient badminton action recognition and evaluation method, which helps to improve the scientific basis and effectiveness of athletes’ training, and also provides a new technical means for sports analysis.

Keywords: Badminton action recognition; MobileNetV3; Improved OpenPose algorithm; DTW algorithm; Joint points

Cite As

Y. Zhong, J. Yang, W. Guo, "Badminton Action Recognition Model Combining Artificial Neural Network
Algorithm and DTW Algorithm", Engineering Intelligent Systems, vol. 33 no. 6, pp. 721-732, 2025.




Downloads

Published

2025-11-01

Issue

Vol. 33 No. 6 (2025): International Journal of Engineering Intelligent Systems

Section

Article

License

The submission of a paper implies that, if accepted for publication, it will not be published elsewhere in the same form, in any language, without the prior consent of the publisher. Before publication, authors are requested to assign copyright to CRL Publishing Ltd. This allows CRL to sanction photocopying, and to authorize the reprinting of issues or volumes according to demand. Authors' traditional rights will not be jeopardized by assigning Copyright in this way, as they retain the right to reuse the material following publication, and to veto third-party publication.