基于树莓派的高精度跌倒检测装置设计与实现

Design and implementation of a high-precision fall detection device based on Raspberry Pi

目的:设计并实现了一款不依赖网络环境、高精度、低迟延的老年人跌倒检测装置。方法:提出一种网格搜索优化的支持向量机跌倒检测算法(IGS-SVM),该方法首先用支持向量机算法对采集到的数据集进行训练,然后再利用改进的网格搜索寻找模型最优参数,最后将模型配置到树莓派终端进行实时跌倒检测。结果:实验表明本跌倒检测系统的准确率为98.06%、敏感度为96.81%,特异度为98.67%。结论:与传统的简单阈值跌倒检测系统相比,本系统能识别的跌倒动作种类较多且跌倒检测准确率更高。

Aims:This paper aims to design and implement a fall detection device with high precision and low latency which does not rely on the network environment.Methods:An improved grid search method was proposed to optimize the support vector machine algorithm()which was used in the fall detection system.Firstly,the collected fall data was trained with IGS-SVM.Then an improved grid search method was used to find the optimal parameters of the fall detection.Finally,the model was deployed to the Raspberry Pi terminal for real-time fall detection.Results:The experimental results showed that the fall detection accuracy of the system was 98.06%.The sensitivity was 96.81%;and the specificity was 98.67%.Conclusions:Compared with the traditional threshold fall detection system,this fall detection system processes more types of activities and has higher fall detection accuracy.