中国农业科学院农业信息研究所 中国农学会
Agricultural Information Institute of CAAS China Association of Agricultural Science Societies
En
Ch
中国农业科学院农业信息研究所 中国农学会 主办

AI检索

订阅

返回顶部
当前位置: 首页 > 文章 > 基于孪生多尺度空洞胶囊网络的黄瓜叶部病害检测方法 江苏农业学报 2023,39 (9) 1827-1833
Position: Home > Articles > Cucumber leaf disease detection based on Siamese multi-scale dilated cap-sule network Jiangsu Journal of Agricultural Sciences 2023,39 (9) 1827-1833

基于孪生多尺度空洞胶囊网络的黄瓜叶部病害检测方法

作  者：
张善文;许新华;齐国红
单  位：
关键词：
黄瓜病害;孪生网络;多尺度空洞卷积;胶囊网络;孪生多尺度空洞胶囊网络
摘  要：
在黄瓜叶部病害检测中,传统方法简单但检测正确率低,难以处理多种多样的病害叶片图像,深度卷积网络的检测正确率高,但依赖于大量训练样本,训练时间长.本研究提出一种基于孪生多尺度空洞胶囊网络(Si-amese multi-scale dilated capsule network,SMSDCNet)的黄瓜叶部病害检测方法,该方法整合了孪生网络、空洞卷积网络和胶囊网络的优势,将多尺度空洞卷积模块Inception引入胶囊网络,作为孪生网络的子网络,构建孪生多尺度空洞胶囊网络模型,提取多尺度判别特征,再进行矢量化处理,最后经动态路由算法得到具有空间位置信息的胶囊向量,进行病害检测与识别.SMSDCNet克服了深度卷积网络需要大量训练样本、训练时间长以及对旋转和仿射变换敏感的问题,并且克服了多尺度卷积网络训练参数较多的问题.在一个较小的黄瓜病害叶片图像数据集上进行试验,病害检测精度达 90%以上.结果表明,该方法能够实现小训练样本集的黄瓜叶部病害检测,为训练样本有限情况下的作物病害检测提供了一种新方法.
译  名：
Cucumber leaf disease detection based on Siamese multi-scale dilated cap-sule network
关键词：
cucumber disease%Siamese network%multi-scale dilated convolution%capsule network%Siamese multi-scale dilated capsule network(SMSDCNet)
摘  要：
In cucumber leaf disease detection,the traditional methods are simple but low detection accuracy,and they are difficult to deal with the various diseased leaf images.Deep convolution neural networks(CNNs)have high detection ac-curacy,but they rely on a large number of training samples,and the training time is long.A cucumber leaf disease detection method based on Siamese multi-scale dilated capsule network(SMSDCNet)was proposed.It integrated the advantages of Sia-mese network,dilated convolution network and capsule network(CapsNet).In SMSDCNet,the multi-scale dilated convolu-tion inception module was introduced into CapsNet to construct the two sub-networks for Siamese multi-scale dilated capsule network model,then the multi-scale discriminant features were extracted and vectorized.Finally,the capsule vector with spa-tial location information was obtained through the dynamic routing algorithm for detecting and recognizing cucumber leaf disea-ses.SMSDCNet overcame the problems of deep convolutional networks that required a large number of training samples,long training time,and sensitivity to rotation and affine transformation,and overcame the problem that multi-scale convolutional networks required more training parameters.Disease detection experiments were conducted on a small cucumber disease leaf image dataset.The detection accuracy was more than 90%.The results showed that the proposed method could detect cucumber leaf disease with small train-ing sample set,which provided a new method for disease detection under the condition of limited training samples.

相似文章

导出引用
计量
文章访问数: 7
HTML全文浏览量: 0
PDF下载量: 0

所属期刊

推荐期刊

友情链接Friendly link
农业信息研究所Agricultural information institute 农业智能知识服务平台Agricultural intelligent knowledge service platform 农业学术服务平台Agricultural Knowledge Service System 中国知网CNKI 维普网VIP 万方WANFANG DATA
联系我们Contact us

电话Tell：010-82109652-808

邮箱E-mail：agrijournal@caas.cn

服务Service
关于我们About us 意见建议Opinions and suggest
版权说明Copyright
单位：中国农业科学院农业信息研究所 中国农学会 © 2017中国农业科学院农业信息研究所 ICP备案号：京ICP备09089781号-46