摘要
针对目前人们纠正体育运动训练误差的手段效率低、误差纠正效果较差的问题,提出利用自动化检测技术设计体育运动训练误差检测系统,对训练过程进行自动测量,从而实现对训练误差的测量和分析.为了降低干扰,该系统采用背景差分法和两帧差分法结合的算法对单目标进行识别,用卷积神经网络算法对多目标进行识别.将召回率和精确率确定为仿真实验的评价指标,选择3个网球公开赛视频进行仿真实验分析,并与其他两种错误动作识别方法进行对比分析.结果表明:该系统进行错误动作识别的召回率和精确率较高,对错误动作的识别效果良好.
Aiming at the problem of low efficiency and poor error correction effect of the means of correcting sports training errors at present,the use of automatic detection technology was proposed to design the sports training error detection system.The training errors were measured and analyzed by the automatic measurement of the training process.To reduce the interference,the sports training error detection system adopted the combination of background difference method and two frame difference method to recognize the single target,and convolutional neural network algorithm to recognize the multi-target.The simulation test of the system was carried out.First,the evaluation indexes of the simulation test were determined as recall rate and accuracy rate,and then three open tennis videos were selected for simulation test analysis,and compared with other two wrong action recognition methods.The results showed that the system had high recall and accuracy rate,and recognition effect of the wrong action was good.
作者
吴明明
WU Ming-ming(Department of Physical Education and Research,Yangen University,Quanzhou 362000,Fujian,China)
出处
《兰州文理学院学报（自然科学版）》
2023年第6期85-90,共6页
Journal of Lanzhou University of Arts and Science(Natural Sciences)
关键词
自动化检测技术
体育训练
误差分析
仿真实验
automatic detection technology
sports training
error analysis
simulation test
