基于YOLOv5的精准农田害虫识别方法研究  

Research on accurate farmland pestidentification method based on YOLOv5

作　　者：孙宏宇 张馨月 宋鑫宇 吴嘉伊 董延华[1] SUN Hong-yu;ZHANG Xin-yue;SONG Xin-yu;WU Jia-yi;DONG Yan-hua(College of Mathematics and Computer,Jilin Normal University,Siping 136000,China)

机构地区：[1]吉林师范大学数学与计算机学院,吉林四平136000

出　　处：《吉林师范大学学报(自然科学版)》2025年第2期113-118,共6页Journal of Jilin Normal University:Natural Science Edition

基　　金：吉林省科技发展计划项目(20220508038RC);国家教育部科技发展中心项目(2022IT096)。

摘　　要：结合目标检测的优势和农业领域亟待解决的害虫预警问题,致力于将目标检测技术应用于病虫害治理,以减少各地农业因病虫害问题造成的经济损失,增加农作物产量.提出以图像识别算法为基础,通过摄像头实时获取图像信息,对其进行预处理和图像增强,以扩充训练集;利用YOLOv5目标检测算法对病虫害信息进行训练,实现了对28种常见农田害虫的识别,并对其中24种准确率达90%以上;完成了对农田病虫害问题的实时监控及信息反馈,在面对病虫害问题时,能够及时治理,从而降低病虫害问题对农作物的影响及损失,提高农作物产量和经济效益.Combining the advantages of object detection with the urgent need for insect pest warning in the agricultural field,we are committed to applying object detection technology to pest control to reduce economic losses caused by pest problems in various regions of agriculture and increase crop yields.We propose using image recognition algorithms to obtain real-time image information through cameras and preprocess and enhance the images to expand the training set.We use the YOLOv5 object detection algorithm to train the pest information and have achieved the recognition of 28 common field pests,with an accuracy rate of 90% or above for 24 of them.We have completed real-time monitoring and information feedback for field pest problems,and can promptly deal with pest problems when they arise,thereby reducing the impact of pest problems on crops and losses and improving crop yields and economic benefits.

关 键 词：目标检测 害虫识别 YOLOv5算法 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]