基于CenterNet的跑步姿态鉴别系统的设计  

Design of running posture identification system based on CenterNet

作　　者：周万珍[1] 袁志鑫 王建霞[1] ZHOU Wanzhen;YUAN Zhixin;WANG Jianxia(School of Information Science and Engineering,Hebei University of Science and Technology,Shijiazhuang,Hebei 050018,China)

机构地区：[1]河北科技大学信息科学与工程学院,河北石家庄050018

出　　处：《河北工业科技》2024年第1期10-16,共7页Hebei Journal of Industrial Science and Technology

基　　金：河北省自然科学基金(F2018208116)。

摘　　要：为了改善目前大众跑步姿势普遍不规范的现状,提出了一种基于CenterNet的跑步姿态鉴别系统。首先,通过截图、拍照的方式自制数据集,并对数据集进行清洗、标注和分析,消除数据无关信息与简化数据。其次,引入多尺度通道注意力机制与添加十字星变形卷积2种方式改进CenterNet算法模型,将动作图像转化为数字信息和特征向量,并以此为基础,利用KNN(K-nearest neighbors)算法对跑步姿态类型进行分类。最后,与经典模型方案进行对比,验证改进CenterNet算法鉴别系统的有效性。结果表明:改进的CenterNet模型的精确率与召回率都有所提升,其参数量与计算量降低。所提算法模型能够对大多数不良姿势作出及时、准确反馈,有效帮助跑步爱好者发现问题,从而改善跑步姿态、提高运动效率、预防伤病。Inorder to improve the current situation that the public running posture is generally not standardized,this paper proposed a running posture identification system based on CenterNet.Firstly,a dataset was created by taking screenshots and photos,and was cleaned,annotated,and analyzed the dataset to eliminate data-independent information and simplify the data.Secondly,the CenterNet algorithm model was improved by introducing multi-scale channel attention mechanism and adding cross star deformation convolution to transform the action image into digital information and feature vector.Then,KNN(K-nearest neighbors)algorithm was used to classify the running posture types.Finally,compared with the classical model scheme,the effectiveness of the improved CenterNet algorithm identification system was verified.The experimental results show that the accuracy and recall rates of the improved CenterNet model are improved,and the parameter quantity and calculation amounts are reduced.It can provide timely and accurate feedback on most bad postures,and can also effectively help running enthusiasts find problems,thereby improving running posture,improving exercise efficiency,and preventing injuries.

关 键 词：计算机图像处理 人体行为识别 跑步姿态 CenterNet 人体关节 注意力机制 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]