基于异常行为分析的疑似病猪实时监测系统设计

Real-time Monitoring System for Suspected Diseased Pigs Based on Abnormal Behavior Analysis

鉴于人工观察猪异常行为识别疑似病猪的方法存在不足,设计了一种基于行为分析的实时监测系统,用于发现疑似病猪。通过部署在猪舍饮食间的长焦镜头和顶部的广角摄像头及从属传感器,对生猪的饮食行为进行24 h监控;通过改进运动目标检测算法和图像识别技术,实现对行为异常的病猪及疑似病猪进行非接触监测,并生成日志和记录数据;预警模块会将图像信息通过网络服务器传送至监控中心,这样可以有效发现猪的异常行为,从而更早地干预和治疗病猪。经过训练模型识别,共发现疑似病猪50只,经过检验,实际患病猪为49只,该系统检测精确度为99%,基本达到了预期的设计目标。

Considering the shortcomings of the method of manually observing abnormal behavior of pigs to identify suspected diseased pigs,a real-time monitoring system based on behavior analysis is designed to detect the suspected diseased pigs.By deploying a telephoto lens and a wide-angle camera at the top of the pigsty food room,as well as a secondary sensor,the dietary behavior of live pigs can be monitored for 24 hours.By improving the motion target detection algorithm and image recognition technology,the non-contact monitoring of diseased pigs and suspected diseased pigs with abnormal behavior is achieved,and logs are generated and data is recorded.The warning module will transmit image information to the monitoring center through a network server,which can effectively detect abnormal behavior of pigs to intervene and treat sick pigs earlier.After training the model recognition,a total of 50 suspected diseased pigs are found.After testing,the actual number of diseased pigs is 49,and the detection accuracy of the system is 99%,basically achieving the expected design goals.