摘要
为了降低体育教学训练错误动作检测的误差率,提升检测效果,研究一种基于深度卷积神经网络的体育教学训练错误动作检测方法。构建多层次的深度卷积神经网络,在输入层中输入未经过特征提取的初始数据,分别经卷积层和池化层处理获取卷积特征图和池化特征图;在卷积层与池化层的中间添加批量归一层,通过批量归一化处理体育教学训练的错误动作样本;在隐含层中重复操作上述步骤并通过设置卷积和池化实现错误动作数据特征的提取,并通过全连接层输出最终的检测结果。实验结果表明,该方法具有较好的体育教学训练错误动作检测效果,检测平均误差率约为0.034%,可准确检测运动人员的各部位训练错误动作。
In order to reduce the error rate of wrong action detection in physical education training and improve the detection effect, a method of wrong action detection in physical education training based on deep convolution neural network is studied in this paper. A multi-level deep convolution neural network is constructed. The initial data without feature extraction is input into the input layer. Convolution feature map and pooling feature map are obtained by convolution layer and pooling layer respectively. A batch normalization layer is added between the convolution layer and pooling layer. The physical education training action samples are processed by batch normalization, and the above steps are repeated in the hidden layer convolution and pooling are set to realize the feature extraction of error action data, and the final detection result is output through the full connection layer. The experimental results show that: this method has a good effect on the detection of wrong movements in physical education training, the average error rate of detection is about 0.034%, and it can accurately detect the wrong movements in various parts of sports personnel training.
作者
刘志鹏
LIU Zhi-peng(Zhangzhou Institute of Technology,Zhangzhou 363000,China)
出处
《三明学院学报》
2021年第3期8-14,共7页
Journal of Sanming University
关键词
深度学习
卷积神经网络
体育教学训练
错误动作检测
特征提取
批量归一化
deep learning
convolution neural network
physical education training
error action detection
Feature extraction
batch normalization
