摘要
农业生产中的玉米病虫害问题日益严重,为提高玉米病虫害的防治精度,进而给农民的生产生活提供更专业的指导,文中提出一种玉米病虫害识别与分级的方法,对常见的5种病虫害进行识别与分级研究。首先,通过对不同类型的轻量级网络对比,选择EfficientNetV2作为特征提取网络;其次,为提高模型的精度,引入DeepViT算法修改网络通道,提升网络的特征提取能力,并引入特征融合网络提高模型的分类精度;最后,将改进后的EDB模型与AlexNet、ResNet、VGG16、DenseNet等网络进行对比。实验结果表明,改进后的模型大小为8.6 MB,玉米病虫害平均识别精度为97.72%,玉米病害分级精度为92.6%,单张图片平均识别时间为24 ms,可实现对玉米病虫害的快速、准确识别,能够为后期玉米的管理提供相应的技术支撑。
The problem of corn pests and diseases in agricultural production is becoming increasingly serious.In order to improve the control accuracy of corn pests and diseases and provide more professional guidance for farmers′production and life,a method of identifying and grading corn pests and diseases is proposed,and the research on identification and classification of five kinds of common diseases and pests is carried out.EfficientNetV2 is selected as the feature extraction network by comparing different types of lightweight networks.The DeepViT algorithm is introduced to modify the network channel and enhance the feature extraction ability of the network,so as to improve the accuracy of the model.A feature fusion network is introduced to improve the classification accuracy of the model.The improved EDB model is compared with AlexNet,ResNet,VGG16,DenseNet and other networks.The experimental results show that the improved model size is 8.6 MB,the average identification accuracy of corn diseases and pests is 97.72%,the grading accuracy of corn diseases is 92.6%,and the average recognition time of a single picture is 24 ms,which can realize the rapid and accurate identification of corn diseases and pests,and provide corresponding technical support for the later-period management of corn growing.
作者
武魁
高丙朋
WU Kui;GAO Bingpeng(School of Electrical Engineering,Xinjiang University,Urumqi 830017,China)
出处
《现代电子技术》
2023年第14期68-74,共7页
Modern Electronics Technique
基金
国家自然科学基金项目(61863033)。