烟草仓储害虫性诱智能监测系统设计与实现  

Design and implementation of an intelligent monitoring system for tobacco storage pests trapped by sex pheromone

作　　者：罗浩伦 李国志 尤彦辰 李彬 吕军[1] 李文冬 姚青[4] LUO Haolun;LI Guozhi;YOU Yanchen;LI Bin;L Jun;LI Wendong;YAO Qing(School of Information Science and Engineering,Zhejiang Sci-Tech University,Hangzhou 310018,China;Guizhou Aerospace Intelligent Agriculture Co.,Ltd.,Guiyang 550000,China;Agricultural Information Center of Changchun City,Jilin Province,Changchun 130051,China;School of Computer Science and Technology,Zhejiang Sci-Tech University,Hangzhou 310018,China)

机构地区：[1]浙江理工大学信息科学与工程学院,杭州310018 [2]贵州航天智慧农业有限公司,贵阳550000 [3]吉林省长春市农业信息中心,长春130051 [4]浙江理工大学计算机科学与技术学院,杭州310018

出　　处：《植物保护》2025年第1期123-131,共9页Plant Protection

基　　金：浙江省科技计划(2022C02004)。

摘　　要：为了实时智能监测烟草仓储害虫,设计并实现了烟草仓储害虫性诱智能监测系统。该系统由基于机器视觉的智能性诱捕器、性诱害虫识别模型、服务器和Web端系统平台组成。智能性诱捕器通过性诱剂将害虫诱集至粘虫板,机器视觉模块每天定时采集一幅粘虫板图像,并通过4G网络将图像上传至服务器。服务器接收到图像后调用性诱害虫识别模型进行害虫的检测与识别,并将检测结果返回到Web客户端。用户可通过系统平台Web端查看诱集的害虫图像和数量。针对粘虫板图像上的性诱害虫烟草甲Lasioderma serricorne和烟草粉螟Ephestia elutella,建立了YOLOX-TP识别模型,在YOLOX的基础上添加了SEnet注意力机制。与Faster-RCNN、YOLOv4、YOLOX检测模型相比,YOLOX-TP平均精确率和平均召回率最高,达到98.97%和97.12%。烟草仓储害虫性诱智能监测系统实现了烟草性诱害虫图像的定时采集、害虫准确检测与计数、结果可视化和可追溯,为烟草仓储害虫防治决策提供依据。For automatically monitoring tobacco storage pests in real-time,an intelligent monitoring system was designed and implemented.The system consists of a machine vision-based intelligent sex pheromone trap,a pest identification model,a server,and a Web-based platform.The intelligent trap attracts pests to a sticky board using sex pheromones,and the machine vision module captures an image of the sticky board at regular intervals every day.The image is then uploaded to the server via a 4G network.Once the server receives the image,it uses the pest identification model to detect and recognize the pests in the image,and returns the results to the network client.Users can view the images and quantities of trapped pests through the network platform.For accurately identifying the Lasioderma serricorne and Ephestia elutella on the sticky board images,a YOLOX-TP recognition model was developed,which incorporates the SEnet attention mechanism into the YOLOX framework.Compared with Faster-RCNN,YOLOv4,and YOLOX models,YOLOX-TP achieved the highest average precision and recall rates,reaching 98.97%and 97.12%,respectively.The intelligent monitoring system for tobacco storage pests achieves real-time image acquisition,accurate pest detection and counting,result visualization,and data traceability,providing a basis for decision-making in the monitoring and controlling tobacco storage pests.

关 键 词：烟草仓储害虫 智能性诱捕器 烟草甲 烟草粉螟 害虫图像 YOLOX-TP模型 

分 类 号：S379.5[农业科学—农产品加工] TP274[农业科学—农艺学] TP183[自动化与计算机技术—检测技术与自动化装置]