Swimming posture recognition and swimsuit auxiliary regulation model combining bionic fluid mechanics and deep learning

View at publisher Summarise with VIRAT
Authors: Dongjun Yang, Ren Xing
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This article focuses on the integration of bionic fluid mechanics and DL (Deep Learning) technology, and is committed to building a swimming posture recognition and swimsuit auxiliary control model. By collecting various images and water flow data of different swimmers, CNN (Convolutional Neural Network) is used to build the attitude recognition model, and the parameters are adjusted to optimize the model after several rounds of training. At the same time, based on the principle of bionic fluid mechanics, the swimsuit control algorithm is established, and the real-time parameter adjustment is realized by feedback control. In the experiment, 50 learners with different levels were selected to collect data in the standard swimming pool to verify the model. The results show that the accuracy of attitude recognition model is 92%, the accuracy of common attitude is over 90%, and the accuracy of subtle attitude is about 85%. Swimwear auxiliary regulation model can significantly improve the swimming efficiency, with a short distance of 25 meters increased by about 8% and a long distance of 100 meters increased by 15%, which is superior to the traditional regulation method of swimwear. The research shows that the fusion model is excellent in swimming posture recognition and swimsuit auxiliary regulation, which provides an effective innovative scheme for swimming training and sports equipment optimization.

Fields of Study & Subjects

Sports Dynamics and Biomechanics Educational Technology and Pedagogy Human Pose and Action Recognition

Publication details

Journal / source
IET conference proceedings.
Volume
2025
Issue
47
Pages
767–773
Publication date
1 March 2026
Language
EN