|本期目录/Table of Contents|

[1]余骥远,高尚兵,李洁,等.基于MS-PLNet和高光谱图像的绿豆叶斑病病级分类[J].江苏农业科学,2023,51(6):178-186.
 Yu Jiyuan,et al.Classification of mung bean leaf spot based on MS-PLNet and hyperspectral images[J].Jiangsu Agricultural Sciences,2023,51(6):178-186.
点击复制

基于MS-PLNet和高光谱图像的绿豆叶斑病病级分类(PDF)
分享到:

《江苏农业科学》[ISSN:1002-1302/CN:32-1214/S]

卷:
第51卷
期数:
2023年第6期
页码:
178-186
栏目:
农业工程与信息技术
出版日期:
2023-03-20

文章信息/Info

Title:
Classification of mung bean leaf spot based on MS-PLNet and hyperspectral images
作者:
余骥远12 高尚兵1 李洁12 陈新2 李士丛2 张浩淼1 袁星星2 唐琪1
1.淮阴工学院,江苏淮安 223001; 2.江苏省农业科学院经济作物研究所,江苏南京 210014
Author(s):
Yu Jiyuanet al
关键词:
绿豆叶斑病图像分类卷积神经网络多尺度特征高光谱图像
Keywords:
-
分类号:
TP391.41;S126
DOI:
-
文献标志码:
A
摘要:
为在绿豆播种过程中提高叶斑病的识别准确率,以实验室获取的高光谱图像为研究对象,提出基于MS-PLNet(multiscale-PlantNet)和高光谱图像的绿豆叶斑病病级分类方法。将绿豆叶片彩色图像通过专业设备转换为高光谱格式,依据病斑大小占叶面积比例,将高光谱图像标注为高抗、抗、中抗、中感、感病5个类别,构成本试验的数据集,然后建立MS-PLNet卷积神经网络模型,该图像分类模型包括图像输入、多尺度特征提取、特征融合、分类等4个阶段。为提高模型对输入不同分辨率图像的适用性,将多尺度图像输入到多尺度特征提取模块,获取分类需要的特征图;为将不同尺度的特征图进行特征融合,通过不同步长的下采样和上采样操作,获取特征融合阶段的输入特征图;为在最终分类阶段获取对分类起作用的特征图,采用通道注意力机制让分类器关注有用的特征,抛弃冗余特征;最后通过使用softmax分类器实现对高光谱图像中高抗、抗、中抗、中感、感病5个绿豆叶斑病病级的分类。通过与已有图像分类方法进行比较,MS-PLNet在训练时获得了最高的验证准确率,训练300次验证准确率为96.8%,训练 1 000 次验证准确率为98.4%;在获得最高验证准确率的同时,训练时间大大缩短,MobileNet-V3在所有与本研究方法进行比较的先进方法中训练时间较短,但是训练300次所花费的时间是MS-PLNet的1.59倍,训练1 000次所花费的时间是MS-PLNet的1.12倍。本研究提出的MS-PLNet模型计算量为0.39 GMac,参数数量为7.75 million,能够高效利用GPU资源。对5个类别的叶斑病图像进行分类时,分类的平均精度达到95.0%,召回率达到99.9%,可以实现叶斑病的高精度识别。本研究所提出的基于MS-PLNet和高光谱图像的分类方法,能够对绿豆叶斑病进行有效的病级分类,同时该方法可以输入多个尺度的图像并且具有较小的参数,可以应用于移动设备实现真实环境下的绿豆叶斑病识别。
Abstract:
-

参考文献/References:

[1]刘慧. 我国绿豆生产现状和发展前景[J]. 农业展望,2012,8(6):36-39.
[2]赵吉平,王彩萍,侯小峰,等. 论绿豆的经济价值及产业化开发利用[J]. 农业科技通讯,2016(5):9-10.
[3]段志龙,赵大雷,刘小进,等. 绿豆常见病害的症状及主要防治措施[J]. 农业科技通讯,2009(6):151-152,160.
[4]刘昌燕,肖炎农,吴小微,等. 绿豆叶斑病病原鉴定及生物学特性研究[J]. 植物保护,2015,41(6):83-87.
[5]Lowe A,Harrison N,French A P. Hyperspectral image analysis techniques for the detection and classification of the early onset of plant disease and stress[J]. Plant Methods,2017,13:80.
[6]吕梦棋,张芮祥,贾浩,等. 基于改进ResNet玉米种子分类方法研究[J]. 中国农机化学报,2021,42(4):92-98.
[7]王美华,吴振鑫,周祖光. 基于注意力改进CBAM的农作物病虫害细粒度识别研究[J]. 农业机械学报,2021,52(4):239-247.
[8]黄英来,艾昕. 改进残差网络在玉米叶片病害图像的分类研究[J]. 计算机工程与应用,2021,57(23):178-184.
[9]闫建伟,张乐伟,赵源,等. 改进RetinaNet的刺梨果实图像识别[J]. 中国农机化学报,2021,42(3):78-83.
[10]赵志焱,杨华,胡志伟,等. 基于TACNN的玉露香梨叶虫害识别[J]. 计算机工程与应用,2021,57(9):176-181.
[11]宋余庆,谢熹,刘哲,等. 基于多层EESP深度学习模型的农作物病虫害识别方法[J]. 农业机械学报,2020,51(8):196-202.
[12]Gao R H,Wang R,Feng L,et al. Dual-branch,efficient,channel attention-based crop disease identification[J]. Computers and Electronics in Agriculture,2021,190:106410.
[13]尚静,黄人帅,张艳,等. 高光谱成像结合模式识别无损检测猕猴桃成熟度[J]. 中国农机化学报,2022,43(8):90-95.
[14]谢鹏尧,富昊伟,唐政,等. 基于RGB图像的冠层尺度水稻叶瘟病斑检测与抗性评估[J]. 浙江大学学报(农业与生命科学版),2021,47(4):415-428.
[15]Hussain A,Balaji S P. Disease classification and detection techniques in rice plant using deep learning[C]//2022 8th international conference on smart structures and systems (ICSSS). Chennai,India:IEEE,2022:1-7.
[16]Bhujel A,Kim N E,Arulmozhi E,et al. A lightweight attention-based convolutional neural networks for tomato leaf disease classification[J]. Agriculture,2022,12(2):228.
[17]刘庆飞,张宏立,王艳玲. 基于深度可分离卷积的实时农业图像逐像素分类研究[J]. 中国农业科学,2018,51(19):3673-3682.
[18]闫壮壮,闫学慧,石嘉,等. 基于深度学习的大豆豆荚类别识别研究[J]. 作物学报,2020,46(11):1771-1779.
[19]杨飚,周芷晴. 基于图像变形网络的小样本图像分类算法研究[J]. 工业控制计算机,2022,35(5):86-88.
[20]Jin X B,Zhang J S,Kong J L,et al. A reversible automatic selection normalization (RASN) deep network for predicting in the smart agriculture system[J]. Agronomy,2022,12(3):591.
[21]Sun N,Li H N. Super resolution reconstruction of images based on interpolation and full convolutional neural network and application in medical fields[J]. IEEE Access,2019,7:186470-186479.
[22]Haase D,Amthor M. Rethinking depthwise separable convolutions:how intra-kernel correlations lead to improved MobileNets[C]//2020 IEEE/CVF conference on computer vision and pattern recognition (CVPR). Seattle,WA,USA:IEEE,2020:14588-14597.
[23]Niu Z Y,Zhong G Q,Yu H. A review on the attention mechanism of deep learning[J]. Neurocomputing,2021,452:48-62.
[24]Hu J,Shen L,Sun G. Squeeze-and-excitation networks[C]//Proceedings of the 2018 IEEE/CVF conference on computer vision and pattern recognition. Salt Lake City,UT,USA:IEEE,2018:7132-7141.
[25]Yang M J,Jiao L C,Liu F,et al. DPFL-nets:deep pyramid feature learning networks for multiscale change detection[J]. IEEE Transactions on Neural Networks and Learning Systems,2022,33(11):6402-6416.
[26]Li J J,Zhao X,Li Y S,et al. Classification of hyperspectral imagery using a new fully convolutional neural network[J]. IEEE Geoscience and Remote Sensing Letters,2018,15(2):292-296.
[27]Sandler M,Howard A,Zhu M,et al. Mobilenetv2:Inverted residuals and linear bottlenecks[C]//Proceedings of the 2018 IEEE/CVF conference on computer vision and pattern recognition. Salt Lake City,UT,USA:IEEE,2018:4510-4520.
[28]Howard A,Sandler M,Chu G,et al. Searching for mobilenetv3[C]//Proceedings of the 2019 IEEE/CVF international conference on computer vision. Seoul,Korea(South):IEEE,2020:1314-1324.
[29]Tan M,Le Q. Efficientnet:rethinking model scaling for convolutional neural networks[EB/OL]. (2019-08-28)[2022-10-01]. https:arxiv.org/abs/1905.11946.
[30]Zhou D Q,Hou Q B,Chen Y P,et al. Rethinking bottleneck structure for efficient mobile network design[M]//Computer Vision-ECCV 2020.Cham:Springer International Publishing,2020:680-697.
[31]He K M,Zhang X Y,Ren S Q,et al. Deep residual learning for image recognition[C]//Proceedings of the 2016 IEEE conference on computer vision and pattern recognition. Las Vegas,NV,USA:IEEE,2016:770-778.
[32]Simonyan K,Zisserman A. Very deep convolutional networks for large-scale image recognition[EB/OL]. (2014-11-04)[2022-10-01]. https://arxiv.org/abs/1409.1556.
[33]Huang G,Liu Z,Van Der Maaten L,et al. Densely connected convolutional networks[C]//Proceedings of the 2017 IEEE conference on computer vision and pattern recognition. Honolulu,HI,USA:IEEE,2017:4700-4708.

相似文献/References:

[1]马晓,邢雪,武青海.基于改进ConvNext的复杂背景下玉米叶片病害分类[J].江苏农业科学,2023,51(19):190.
 Ma Xiao,et al.Maize leaf disease classification under complex background based on improved ConvNext[J].Jiangsu Agricultural Sciences,2023,51(6):190.
[2]谭彬,蔡健荣,许骞,等.基于注意力机制改进卷积神经网络的柑橘病虫害识别[J].江苏农业科学,2024,52(8):176.
 Tan Bin,et al.Recognition of citrus pests and diseases based on attention mechanism improved convolutional neural networks[J].Jiangsu Agricultural Sciences,2024,52(6):176.
[3]王晶,崔艳荣.基于改进MobileNet v3-Small模型的草莓病害识别方法[J].江苏农业科学,2024,52(10):225.
 Wang Jing,et al.Strawberry disease identification method based on improved MobileNet v3-Small model[J].Jiangsu Agricultural Sciences,2024,52(6):225.
[4]张澳雪,崔艳荣,李素若,等.基于改进RegNet网络的玉米叶片病害识别研究[J].江苏农业科学,2024,52(11):216.
 Zhang Aoxue,et al.Identification of maize leaf diseases based on improved RegNet network[J].Jiangsu Agricultural Sciences,2024,52(6):216.
[5]王浩宇,崔艳荣.基于改进ShuffleNet v2模型的苹果叶片病害识别方法[J].江苏农业科学,2024,52(13):214.
 Wang Haoyu,et al.Apple leaf disease identification method based on improved ShuffleNet v2 model[J].Jiangsu Agricultural Sciences,2024,52(6):214.
[6]梁倩倩,陈勇,崔艳荣.基于改进轻量化网络MobileViT的苹果叶片病虫害识别方法[J].江苏农业科学,2024,52(14):222.
 Liang Qianqian,et al.An apple leaf pest identification method based on improved lightweight network MobileViT[J].Jiangsu Agricultural Sciences,2024,52(6):222.
[7]戴硕,白涛,李东亚,等.基于知识蒸馏及改进ShuffleNet v2的棉花病虫害识别方法[J].江苏农业科学,2024,52(15):222.
 Dai Shuo,et al.Recognition of cotton pests and diseases based on knowledge distillation and improved ShuffleNet v2[J].Jiangsu Agricultural Sciences,2024,52(6):222.
[8]严露露,朱赞彬,冯世杰,等.基于改进FixMatch算法的半监督番茄病虫害识别[J].江苏农业科学,2024,52(20):244.
 Yan Lulu,et al.Semi-supervised identification of tomato diseases and pests based on improved FixMatch algorithm[J].Jiangsu Agricultural Sciences,2024,52(6):244.
[9]董天亮,李佳,马晓,等.基于SC-ConvNeXt网络模型的小麦叶片病害识别方法[J].江苏农业科学,2025,53(5):129.
 Dong Tianliang,et al.A wheat leaf disease recognition method based on SC-ConvNeXt network model[J].Jiangsu Agricultural Sciences,2025,53(6):129.

备注/Memo

备注/Memo:
收稿日期:2022-10-14
基金项目:国家自然科学基金面上项目(编号:62076107);国家重点研发计划(编号:2018YFB1004904);江苏省高校自然科学研究重大项目(编号:18KJA520001);江苏省产学研合作项目(编号:BY2022334)
作者简介:余骥远(1997—),男,湖北襄阳人,硕士研究生,研究方向为图像处理。E-mail:421226718@qq.com。
通信作者:高尚兵,博士,教授,主要研究方向深度学习、图像处理。E-mail:11060036@hyit.edu.cn。
更新日期/Last Update: 2023-03-20