Swimming motion capture and posture analysis algorithm based on inertial sensor and multi-modal motion features

Swimming, as a comprehensive aerobic exercise, not only delves into the mechanics of swimming, but also facilitates the prevention of sports injuries through motion capture and posture analysis. To capture swimming motions, an inertial sensor is used and a swimming motion acquisition system based on a micro-electro-mechanical systems is designed. In response to the errors in the inertial sensor, a zero velocity update is adopted for processing. To perform posture analysis on swimming motions, multi-modal features including signal features, statistical features, and posture features are collected. In addition, the study adopts adaptive differential evolution algorithm and introduces an extreme learning machine to optimize it, forming the final phase segmentation algorithm. The results showed that the data collected in the study could describe the characteristics of different swimming strokes. The root mean square errors of the pan zero posture correction algorithm in roll angle, pitch angle, and heading angle were 0.0925, 0.0912, and 0.0457, respectively, indicating good posture solution performance. On the upper left arm, the average absolute error between the proposed collection system and the OptiTrack optical motion capture system (true value) was 0.0738. The maximum accuracy of the phase segmentation model in swimming posture recognition was 99.73%, with an average root mean square error of 0.5137. The recognition time for the arm stroke phase in breaststroke was 32.3 ms, and the recognition sensitivity for the butterfly leg phase was 98.72%. The data collection system and designed phase segmentation model have good performance, which provide technical support for swimming posture analysis to improve the swimming proficiency of swimming enthusiasts. The research contribution lies in successfully capturing swimming motions and analyzing postures, which provides a scientific basis for training guidance for swimmers.

Authors

Institutions

Publication Details

Journal
Discover Applied Sciences
Published
2026-09-15
DOI
https://doi.org/10.1007/s42452-026-09490-4
Primary Topic
Advanced Technologies in Various Fields
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
article

Swimming motion capture and posture analysis algorithm based on inertial sensor and multi-modal motion features

Ran Li
Discover Applied Sciences
Advanced Technologies in Various Fields
article

Swimming motion capture and posture analysis algorithm based on inertial sensor and multi-modal motion features

Ran Li
article en

Abstract

Swimming, as a comprehensive aerobic exercise, not only delves into the mechanics of swimming, but also facilitates the prevention of sports injuries through motion capture and posture analysis. To capture swimming motions, an inertial sensor is used and a swimming motion acquisition system based on a micro-electro-mechanical systems is designed. In response to the errors in the inertial sensor, a zero velocity update is adopted for processing. To perform posture analysis on swimming motions, multi-modal features including signal features, statistical features, and posture features are collected. In addition, the study adopts adaptive differential evolution algorithm and introduces an extreme learning machine to optimize it, forming the final phase segmentation algorithm. The results showed that the data collected in the study could describe the characteristics of different swimming strokes. The root mean square errors of the pan zero posture correction algorithm in roll angle, pitch angle, and heading angle were 0.0925, 0.0912, and 0.0457, respectively, indicating good posture solution performance. On the upper left arm, the average absolute error between the proposed collection system and the OptiTrack optical motion capture system (true value) was 0.0738. The maximum accuracy of the phase segmentation model in swimming posture recognition was 99.73%, with an average root mean square error of 0.5137. The recognition time for the arm stroke phase in breaststroke was 32.3 ms, and the recognition sensitivity for the butterfly leg phase was 98.72%. The data collection system and designed phase segmentation model have good performance, which provide technical support for swimming posture analysis to improve the swimming proficiency of swimming enthusiasts. The research contribution lies in successfully capturing swimming motions and analyzing postures, which provides a scientific basis for training guidance for swimmers.

Discover Applied Sciences
Chongqing Vocational and Technical University of Mechatronics (CN), Chongqing Vocational Institute of Engineering (CN)
Life below water
Openalex Percentile: Top 8%
Advanced Technologies in Various Fields
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.