KINECT BASED RECOGNITION AND DETECTION OF FITNESS QIGONG MOVEMENTS

作     者：Fan, Zhehua Sun, Keshuang 

作者机构：Putian Univ Coll Phys Educ 1133 Xueyuan Rd Putian 351100 Peoples R China Fuzhou Univ Dept Phys Educ & Res 2 Wulong Jiangbei Ave Fuzhou 350108 Peoples R China 

出 版 物：《INTERNATIONAL JOURNAL OF INNOVATIVE COMPUTING INFORMATION AND CONTROL》 (Int. J. Innov. Comput. Inf. Control)

年 卷 期：2024年第20卷第6期

页      面：1837-1850页

核心收录：

学科分类：08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：Fujian Province Higher Education Teaching Research Project  Innovation and Practice of Cultivating Applied Talents in Physical Education Based on OBE Concept [FBJY20230213] 

主　　题：Kinect Action recognition DTW algorithm Artificial neural network Stat- ic K-means 

摘      要：The continuous advancement of computer technology and sensory games has provided players with a richer experience, but in the field of sports and fitness, especially in the teaching and training of fitness qigong, how to accurately identify and correct movements remains a major challenge. Therefore, this study focuses on the recognition of human movements and proposes a Kinect based recognition and detection of fitness qigong movements. The purpose is to more accurately grasp the data of body scans and body contours of athletes to correct their incorrect body movements, thereby reducing the difficulty of movement and overcoming the difficulties of traditional sports training. Firstly, the pose recognition method based on the static K-means algorithm is used to calculate and recognize human skeletal joints. Secondly, the dynamic time warping algorithm is combined to recognize and detect gymnastics posture features, and experimental analysis is conducted on the effectiveness and applicability of recognition and detection techniques. The experiment shows that the recognition accuracy for stretching and chest expansion movements is 247 and 253, respectively, while the recognition accuracy for these two movements in the preprocessed dataset is improved to 259 and 263, respectively. It can be seen that this study not only improves the accuracy of motion recognition, but also promotes the development of personalized fitness guidance. The method proposed by the research institute provides coaches with real-time feedback tools, optimizes athlete training plans, and achieves the improvement of athlete technical level and sports effectiveness. Not only did it help popularize the scientific fitness concept, but it also improved the overall teaching quality and efficiency of fitness qigong.