%0 Journal Article
%T 通过轻量级人体动作识别实现老年人安全实时监控<br>Through Lightweight Human Movement Recognition to Realize the Security of the Elderly Real-Time Monitoring
%A 吴嘉轩
%A 申奇
%J Software Engineering and Applications
%P 198-208
%@ 2325-2278
%D 2023
%I Hans Publishing
%R 10.12677/SEA.2023.122020
%X 本文旨在探讨如何通过轻量级人体动作识别技术实现老年人安全实时监控。老年人安全问题一直是一个备受关注的问题，尤其是在现代化社会，许多老年人居住在独自生活的环境中，缺乏及时的照顾和照料。为了及时监控老年人的安全情况，本文提出了一种面向老年人实时监控系统的轻量级人体动作识别算法。本文使用卷积神经网络与具有长短时记忆的循环神经网络的组合架构，结合了卷积神经网络在图像特征提取过程中的优越性能，以及长短时记忆神经网络对时序数据处理过程的特点，并进行了详细的实验验证，实验结果表明本文提出的轻量级人体动作识别算法具有显著的优势。<br />
The purpose of this paper is to explore how to achieve real-time monitoring of elderly safety through lightweight human motion recognition technology. The safety of the elderly has always been a matter of great concern, especially in modern society where many elderly people live in solitary environments and lack timely care and attention. In order to monitor the safety of the elderly in a timely manner, this paper proposes a lightweight human action recognition algorithm for a real-time monitoring system for the elderly. This paper uses a combined architecture of convolutional neural network and recurrent neural network with long and short term memory, combines the superior performance of convolutional neural network in image feature extraction process and the characteristics of long and short term memory neural network for temporal data processing process, and conducts detailed experimental validation, the experimental results show that the lightweight human action recognition algorithm proposed in this paper has significant advantages.
%K 老年人安全监控，轻量级人体动作识别，长短时记忆循环神经网络，动作识别，智能监控<br>Security Monitoring of the Elderly
%K Lightweight Human Movement Recognition
%K Long Short Term Memory Recurrent Neural Network
%K Action Recognition
%K Intelligent Monitoring
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=63495