摘要
2016年8月,国际奥委会正式宣布冲浪运动成为东京奥运会正式比赛项目,给冲浪运动的发展带来了巨大机遇。动作识别作为人工智能技术赋能体育的重要应用,通过在时域上跟踪关键点运动以记录运动员的运动动作,并将其转换为数学方式表达运动过程,对于竞技训练和全民健身均具有重要意义。本研究采用惯性测量单元(IMU)采集数据样本,并通过支持向量机(SVM)模型和隐马尔可夫(HMM)模型两种机器学习方法对冲浪运动员动作识别的结果进行了比对。结果表明:SVM模型(精度83.4%)与HMM模型(精度91.4%)均能够有效地进行冲浪运动员的动作识别,但与SVM模型相比,HMM模型的分类精度更高。该算法实现了对冲浪过程中佩戴的IMU输入数据进行时变运动分类,为同步视频关联的非传统体育运动项目动作识别提供了一种新方法和新应用。
In August 2016,the International Olympic Committee officially announced that surfing has become an official event of the Tokyo Olympics,which has brought great opportunities for the development of surfing.As an important application of artificial intelligence technology to empower sports,motion recognition records the athlete s movement process by tracking key points in the time domain,and converts it into a mathematical way to express the movement process,which is of great significance for competitive training and national fitness.The research uses inertial measurement unit(IMU)to collect data samples,and compares the results of surfer action recognition by using support vector machine(SVM)and hidden Markov(HMM)machine learning methods.The results show that both SVM model(accuracy 83.4%)and HMM model(accuracy 91.4%)can effectively recognize surfer actions,but compared with SVM model,HMM model has higher classification accuracy.This algorithm realizes the time-varying motion classification of IMU input data worn during surfing,and provides a new method and new application for the action recognition of non-traditional sports events associated with synchronized videos.
作者
杨铃春
王翔宇
尚志强
YANG Ling-chun;WANG Xiang-yu;SHANG Zhi-qiang(School of Physical Education,Changsha University,Changsha,Hunan 410022,China)
出处
《体育研究与教育》
2024年第5期68-73,共6页
Sports Research and Education
基金
湖南省社科基金项目(21YBQ097)。
