摘要
体育运动动作识别是计算机视觉研究的热点问题.将计算机虚拟现实技术应用到大学体育教学之中,将获取的学生运动动作作为输入和虚拟场景进行交互.该方法采用的是一种半监督框架下实现的算法,首先基于Q统计量进行虚拟与现实差异性度量选择算法来挑选出自适应学习能力较强的体育生,然后利用分类器近邻置信度公式从没有被标记的体育生中选择具有较高置信度水平的学生,并将其归入到已标记的体育生之中,推动模型泛化能力的提高.实验结果表明,该方法可有效辅助体育教学活动,为教学结果提供客观、有效的数据分析.
Sports action recognition is a hot research issue in computer vision.In this paper,the computer virtual reality technology is applied to university physical education,and the acquired student movement is used as input to interact with the virtual scene.The method uses a semi-supervised framework to achieve the algorithm.Firstly,sports majors with sound self-learning abilities are picked out based on the selection algorithm which is achieved by the differences between statistics of the virtual and of the reality.Then,using the confidence coefficient of the classifier committee,students with higher confidence level are selected from unmarked PE students and put into labeled PE students to promote the enhancement of model generalization ability.Experimental results show that this method can effectively assist physical education activities and provide objective and effective data analysis for teaching results.
出处
《兰州文理学院学报(自然科学版)》
2018年第1期120-124,共5页
Journal of Lanzhou University of Arts and Science(Natural Sciences)
基金
安徽高校人文社会科学研究项目(SK2017A0365)
安徽省体育社会科学研究项目(ASS2016112)
关键词
体育运动动作识别
半监督训练
自适应学习
虚拟现实技术
motion recognition
semi-supervised training
adaptive learning
virtual reality technology