SD-YOLOv8: Automated Motion Detection System for Aerobics Students

Existing object detection algorithms are often constrained by the high computational costs associated with large network structures in practical applications. To facilitate the development of intelligent physical education classrooms, particularly for aerobic exercise student movement recognition, w...

Full description

Saved in:
Bibliographic Details
Main Authors: Lian Tang, Ya Li, Qing Du, Jincheng Liang
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11069293/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849704080799170560
author Lian Tang
Ya Li
Qing Du
Jincheng Liang
author_facet Lian Tang
Ya Li
Qing Du
Jincheng Liang
author_sort Lian Tang
collection DOAJ
description Existing object detection algorithms are often constrained by the high computational costs associated with large network structures in practical applications. To facilitate the development of intelligent physical education classrooms, particularly for aerobic exercise student movement recognition, we propose a lightweight student action recognition system based on SD-YOLOv8. First, we constructed a specialized dataset by capturing and annotating aerobic exercise student movement images. Then, based on the YOLOv8 network architecture, we implemented two key improvements: 1) introducing Selective Kernel Networks (SKNet) in the multi-scale feature fusion layer to enhance local feature capture capabilities; 2) replacing the original detection head with Dynamic Head to effectively reduce the computational burden while improving detection accuracy and efficiency. Finally, we developed a user-friendly graphical interface using the PyQt framework, enabling visual deployment and real-time interaction with the model. Experimental results demonstrate that the improved SD-YOLOv8 achieves 97.6%, mean Average Precision (mAP) on the aerobic exercise student behavior recognition dataset, representing a 3.5% improvement over the original model. The system shows significant advantages in enhancing action recognition accuracy and real-time feedback, providing robust technical support for physical education instruction.
format Article
id doaj-art-c9cef5357feb4fd5a1d272876ccbb020
institution DOAJ
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-c9cef5357feb4fd5a1d272876ccbb0202025-08-20T03:16:56ZengIEEEIEEE Access2169-35362025-01-011311805511806510.1109/ACCESS.2025.358556111069293SD-YOLOv8: Automated Motion Detection System for Aerobics StudentsLian Tang0https://orcid.org/0009-0000-6054-4171Ya Li1Qing Du2https://orcid.org/0009-0000-2541-0442Jincheng Liang3https://orcid.org/0009-0004-7046-7199School of Sports Science and Engineering, Hunan Engineering University, Xiangtan, ChinaSchool of Electrical Information Engineering, Hunan Engineering University, Xiangtan, ChinaSchool of Resources Environment and Safety Engineering, University of South China, Hengyang, ChinaSchool of Resources Environment and Safety Engineering, University of South China, Hengyang, ChinaExisting object detection algorithms are often constrained by the high computational costs associated with large network structures in practical applications. To facilitate the development of intelligent physical education classrooms, particularly for aerobic exercise student movement recognition, we propose a lightweight student action recognition system based on SD-YOLOv8. First, we constructed a specialized dataset by capturing and annotating aerobic exercise student movement images. Then, based on the YOLOv8 network architecture, we implemented two key improvements: 1) introducing Selective Kernel Networks (SKNet) in the multi-scale feature fusion layer to enhance local feature capture capabilities; 2) replacing the original detection head with Dynamic Head to effectively reduce the computational burden while improving detection accuracy and efficiency. Finally, we developed a user-friendly graphical interface using the PyQt framework, enabling visual deployment and real-time interaction with the model. Experimental results demonstrate that the improved SD-YOLOv8 achieves 97.6%, mean Average Precision (mAP) on the aerobic exercise student behavior recognition dataset, representing a 3.5% improvement over the original model. The system shows significant advantages in enhancing action recognition accuracy and real-time feedback, providing robust technical support for physical education instruction.https://ieeexplore.ieee.org/document/11069293/Intelligent sportsintelligent classroommotion recognitionYOLOv8
spellingShingle Lian Tang
Ya Li
Qing Du
Jincheng Liang
SD-YOLOv8: Automated Motion Detection System for Aerobics Students
IEEE Access
Intelligent sports
intelligent classroom
motion recognition
YOLOv8
title SD-YOLOv8: Automated Motion Detection System for Aerobics Students
title_full SD-YOLOv8: Automated Motion Detection System for Aerobics Students
title_fullStr SD-YOLOv8: Automated Motion Detection System for Aerobics Students
title_full_unstemmed SD-YOLOv8: Automated Motion Detection System for Aerobics Students
title_short SD-YOLOv8: Automated Motion Detection System for Aerobics Students
title_sort sd yolov8 automated motion detection system for aerobics students
topic Intelligent sports
intelligent classroom
motion recognition
YOLOv8
url https://ieeexplore.ieee.org/document/11069293/
work_keys_str_mv AT liantang sdyolov8automatedmotiondetectionsystemforaerobicsstudents
AT yali sdyolov8automatedmotiondetectionsystemforaerobicsstudents
AT qingdu sdyolov8automatedmotiondetectionsystemforaerobicsstudents
AT jinchengliang sdyolov8automatedmotiondetectionsystemforaerobicsstudents