基于瑞芯微RK3566平台和Yolov8n的输液监测系统  

An Infusion Monitoring System Based on Rockchip RK3566 Platform and Yolov8n

作　　者：杜泽明 Du Zeming(Guangdong Meikong Smart Building Co.,Ltd.,Foshan,Guangdong 528308,China)

机构地区：[1]广东美控智慧建筑有限公司,广东佛山528308

出　　处：《计算机时代》2025年第4期41-45,51,共6页Computer Era

摘　　要：随着智慧病房和智慧医疗的快速发展,非侵入式输液监测的需求日益增长。传统的视觉监测技术依赖于高算力服务器,还存在隐私泄露的风险。本文提出一种基于YOLOv8n模型和瑞芯微RK3566平台的非侵入式边缘端输液检测系统,结合模型量化技术,能够在边缘端计算设备上实现本地推理,结合滑动窗口线性回归算法,系统可以预测输液时间并发出告警。实验表明,该系统能够在边缘设备上高效运行,具备实时性与准确性,可以减少医护人员工作负担的同时保障患者隐私安全。With the rapid development of smart wards and smart healthcare,the demand for non-contact infusion monitoring is increasing.Traditional vision-based monitoring technology relies on high-performance servers,posing a risk to patient privacy.This paper proposes a non-invasive edge-based infusion detection system based on the YOLOv8n model and Rockchip RK3566 platform,combining with model quantization technology,enabling local inference on edge computing devices.The system,incorporating a sliding window linear regression algorithm,can predict infusion time and issue alerts.Experimental results show that the system runs efficiently on edge devices,ensuring real-time performance and accuracy,reducing the workload of medical staff while safeguarding patient privacy.

关 键 词：边缘端计算 RK3566 YOLOv8n 模型量化 滑动窗口 线性回归 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]