基于改进YOLOv7的林业害虫检测分类方法

打开文本图片集

关键词：林业害虫；目标检测；YOLOv7；GhostConv；SE注意力；CoordConv 中图分类号：TP391.4；S763.3 文献标识码：A DOI：10.7525/j.issn.1006-8023.2026.01.015

Forest Pest Detection and Classification Methods Based on Improved YOLOv7

ZHU Qiangjun1，LIU Chenxin²，WANG Yang3* （1.ScholofComputingScience，WuhuUniversity，Wuhu2410o，China;；2.ScholofElectronic&InformationTechnology，Wuh University，Wuhu2410o，China；3.SchoolofComputerandInformation，AnhuiNormalUniversity，Wuhu2410o，China）

Abstract：Inorderto improve the accuracyof forestpest identification，a forest pestdetection model（GhostConvand SE atention enhanced YOLOv7，GS-YOLOv7） based onthe improved YOLOv7 is proposed.Firstly，the model replaced the traditionalconvolution in the backbone network with GhostConv lightweightconvolution to reduce the numberof parameters in model operation and improve the model efficiency.Secondly，byadding the squeeze excitation （SE）attention module，theabilitytoextracttheedgesofpestimages withinsignificantfeatures was enhanced，therebyfurtherimprovingthe feature extractionabilityof the network.Thirdly，thecontent aware reassemblyoffeatures （CARAFE）lightweight operator was used toreplace thetraditional upsampling method to improve the qualityoffeature reconstruction， solvethescale mismatch problem，and enhance thedetection performance.Finall，the cordinate convolution（CoordConv）modulewas introduced into the Necknetwork，and its position information was utilizedto solvethe problemof inaccuratetarget positioning and improve the model'ssensitivityto spatial positions andits generalizationability.Experimentswere conducted on six common pest and disease datasets，the precision of the GS-YOLOv7 model reached 93. 15% ，and the mean average precision at an intersection over union threshold of O.5 reached 93.29% . Compared with the original model， the precision and mean average precision increased by 6.50% and 2.09% ，respectively.The number of parameters and the model size decreased to 1.9×107 units and 38.17 MB，representing a reduction of 51.4% （204 and 46.53% ，respectively，compared to the original model. Results indicate that the GS-YOLOv7 model demonstrates significant performance improvements overthe original model，confirming the effectiveness of the modelmodifications. Keywords：Forest pests；object detection； YOLOv7; GhostConv;squeeze excitation atention； CoordConv

0 引言

中国林业遭受生物胁迫的有害生物种类有8000多种，经常造成危害的有200多种[1]。（剩余13104字）