中国农机化学报2026，Vol.47Issue(2)：210-216,225,8.DOI:10.13733/j.jcam.issn.2095-5553.2026.02.028
基于改进RT—DETR的棉田昆虫检测算法
Insect detection algorithm for cotton fields based on improved RT—DETR
摘要
Abstract
Given the persistent challenges of limited accuracy,frequent missed detections,and false positives in cotton field insect detection,this study proposed an improved cotton field insect detection algorithm based on the RT—DETR framework.To address these issues,several key enhancements were introduced.First,WTConv was used to replace the second conventional convolution layer in the residual block.This modification significantly expanded the model's receptive field while keeping the number of trainable parameters low,thereby enhancing the model's ability to detect small targets effectively.Next,a dual-branch M2SA module was incorporated to extract both global features and channel information.This improved the model's understanding of complex field environments and boosted its accuracy in detecting small insects.Additionally,during the cross-scale feature fusion stage,a Small Target Optimization Pyramid(STOP)was designed to efficiently capture and integrate both global and local features,further improving the detection of small targets.Experimental results showed that the improved RT—DETR model achieved a mean Average Precision(mAP)of 95.4%,which was 8.9 percentage points higher than that of the original RT—DETR model.Furthermore,the enhanced model required only 12.1 million parameters and 42 G of floating point operations per second,representing a 36%reduction in parameters and a 26%reduction in computational load compared to the original version.In summary,the improved RT—DETR model significantly improved the accuracy and efficiency of insect detection in cotton fields,offering a practical precise solution for pest monitoring and management in agricultural applications.关键词
棉田昆虫/目标检测/RT—DETR/小波变换卷积Key words
cotton field insects/target detection/RT—DETR/wavelet transform convolution分类
农业科技引用本文复制引用
Chen Kang,Chen Lin..基于改进RT—DETR的棉田昆虫检测算法[J].中国农机化学报,2026,47(2):210-216,225,8.基金项目
中国高校产学研创新基金(2020ITA03012) （2020ITA03012）
