[1]蒋东山,刘金洋,张浩淼,等.基于CNN和Transformer的绿豆干旱胁迫识别模型[J].江苏农业学报,2025,(01):87-100.[doi:doi:10.3969/j.issn.1000-4440.2025.01.011]
　JIANG Dongshan,LIU Jinyang,ZHANG Haomiao,et al.Drought stress recognition model of mung bean based on CNN and Transformer[J].,2025,(01):87-100.[doi:doi:10.3969/j.issn.1000-4440.2025.01.011]
点击复制

基于CNN和Transformer的绿豆干旱胁迫识别模型()
分享到：

江苏农业学报[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2025年01期
页码:
87-100
栏目:
农业信息工程
出版日期:
2025-01-31

文章信息/Info

Title:
Drought stress recognition model of mung bean based on CNN and Transformer
作者:
蒋东山1; 2; 3; 刘金洋2; 3; 张浩淼1; 2; 李士丛2; 罗仔秋1; 余骥远1; 2; 李洁1; 2; 陈新2; 袁星星2; 4; 高尚兵1; 4
(1.淮阴工学院计算机与软件工程学院,江苏淮安223003；2.江苏省农业科学院经济作物研究所,江苏南京210014)
Author(s):
JIANG Dongshan1; 2; 3; LIU Jinyang2; 3; ZHANG Haomiao1; 2; LI Shicong2; LUO Zaiqiu1; YU Jiyuan1; 2; LI Jie1; 2; CHEN Xin2; YUAN Xingxing2; 4; GAO Shangbing1; 4
(1.Faculty of Computer and Software Engineering, Huaiyin Institute of Technology, Huai’an 223003, China;2.Institute of Economic Crops, Jiangsu Academy of Agricultural Sciences, Nanjing 210014, China)
关键词:
绿豆; 干旱胁迫; 卷积神经网络; 转换器; 图像识别; 叶绿素荧光图像
Keywords:
mung bean; drought stress; convolutional neural network; transformer; image recognition; chlorophyll fluorescence image
分类号:
S522
DOI:
doi:10.3969/j.issn.1000-4440.2025.01.011
文献标志码:
A
摘要:
为了解决传统绿豆干旱胁迫识别方法存在识别率低、时效性差的问题,本研究建立了基于卷积神经网络(CNN)和转换器(Transformer)的绿豆干旱胁迫识别模型Mungbean-droughtNet。该模型采用双分支结构,利用全局特征提取模块(GFEM)分支和局部特征提取模块(LFEM)分支分别从输入图像提取局部特征和全局特征。最后利用多层感知器(MLP)模块将局部特征和全局特征进行融合,实现分类。在实际数据分析中,共采集14 536张干旱胁迫下的绿豆叶绿素荧光图像,分为HR、R、MR、S、HS和对照6类。利用Mungbean-droughtNet模型对叶绿素荧光图像数据集进行分析,结果表明,Mungbean-droughtNet模型对测试集中叶绿素荧光图像的平均识别准确率为95.57%,平均精度为98.18%,平均召回率为98.40%,平均F1分数为98.28%。和目前先进模型EfficientNetV2和Swin Transformer相比,Mungbean-droughtNet模型准确率分别提高了3.56个百分点和2.62个百分点,表现出更强的鲁棒性和更好的识别效果。本研究结果为绿豆干旱胁迫研究和耐旱基因挖掘提供了基础。
Abstract:
To address the issues of low recognition rate and poor timeliness in traditional methods for identifying drought stress in mung beans, this study established a mung bean drought stress recognition model named Mungbean-droughtNet based on convolutional neural network (CNN) and transformer. The model employed a dual-branch structure, utilized the global feature extraction module (GFEM) branch and the local feature extraction module (LFEM) branch to extract local and global features from the input images, respectively. Finally, multilayer perceptron (MLP) module was used to fuse the local and global features for classification. In the actual data analysis, a total of 14 536 chlorophyll fluorescence images of mung beans under drought stress were collected and classified into six categories, including HR, R, MR, S, HS and the control group. The Mungbean-droughtNet model was applied to analyze the chlorophyll fluorescence image dataset, the results showed that the Mungbean-droughtNet model achieved an average recognition accuracy of 95.57%, an average precision of 98.18%, an average recall ratio of 98.40%, and an average F1-score of 98.28% for the chlorophyll fluorescence images in the test set. Compared with the current advanced models EfficientNetV2 and Swin Transformer, the accuracy of the Mungbean-droughtNet model increased by 3.56 percentage points and 2.62 percentage points, respectively, demonstrating stronger robustness and better recognition performance. This study provides a foundation for research on mung bean drought stress and the excavation of drought-resistant genes.

参考文献/References:

[1]李茉莉,孙桂华,赵阳. 小杂粮作物的开发价值及战略意义[J]. 园艺与种苗,2005(2):123-124.
[2]郑卓杰. 中国食用豆类学[M]. 北京:中国农业出版社,1997.
[3]《气候变化国家评估报告》编写委员会. 气候变化国家评估报告[M]. 北京:科学出版社,2007.
[4]翟盘茂,王萃萃,李威. 极端降水事件变化的观测研究[J]. 气候变化研究进展,2007,3(3):144-148.
[5]OCHSNER T E, COSH M H, CUENCA R H, et al. State of the art in large-scale soil moisture monitoring[J]. Soil Science Society of America Journal,2013,77(6):1888-1919.
[6]金祎婷,刘文辉,刘凯强,等. 全生育期干旱胁迫对‘青燕1号’燕麦叶绿素荧光参数的影响[J]. 草业学报,2022,31(6): 112-126.
[7]岳焕然,李茂松,安江勇.基于颜色和纹理特征的玉米干旱识别[J]. 中国农学通报,2018,34(24):18-28.
[8]NAIK H S, ZHANG J, LOFQUIST A, et al. A real-time phenotyping framework using machine learning for plant stress severity rating in soybean[J]. Plant Methods,2017,13:1-12.
[9]龙燕,马敏娟,王英允,等. 利用叶绿素荧光动力学参数识别苗期番茄干旱胁迫状态[J]. 农业工程学报,2021,37(11):172-179.
[10]ZHOU C Y, LE J, HUA D X, et al. Imaging analysis of chlorophyll fluorescence induction for monitoring plant water and nitrogen treatments[J]. Measurement,2018,136(6):1-9.
[11]AN J, LI W, LI M, et al. Identification and classification of maize drought stress using deep convolutional neural network[J]. Symmetry,2019,11(2):256.
[12]XIE W, WEI S, ZHENG Z, et al. A cnn-based lightweight ensemble model for detecting defective carrots[J]. Biosystems Engineering,2021,208(2):287-299.
[13]KAMILARIS A, PRENAFETA-BOLD F X. Deep learning in agriculture:a survey[J]. Computers and Electronics in Agriculture,2018,147:70-90.
[14]ZHANG X, XUN Y, CHEN Y. Automated identification of citrus diseases in orchards using deep learning[J]. Biosystems Engineering,2022,223:249-258.
[15]ABADE A D S, PORTO L F, FERREIRA P A, et al. NemaNet:a convolutional neural network model for identification of soybean nematodes[J]. Biosystems Engineering,2022,213:39-62.
[16]KOZOWSKI M, GRECKI P, SZCZYPINSKI P M. Varietal classification of barley by convolutional neural networks[J]. Biosystems Engineering,2019,184:155-165.
[17]LECUN Y, BENGIO Y, HINTON G. Deep learning[J]. Nature,2015,521(7553):436-444.
[18]赵奇慧 ,李莉,张淼,等. 基于迁移学习的温室番茄叶片水分胁迫诊断方法[J]. 农业机械学报,51(增刊1):340-347.
[19]OCHSNER T E, COSH M H, CUENCA R H, et al. State of the art in large-scale soil moisture monitoring[J]. Soil Science Society of America Journal,2013,77(6):1888-1919.
[20]CHANDEL S N, CHAKRABORTY K S, RAJWADE A Y, et al. Identifying crop water stress using deeplearning models[J]. Neural Computing and Applications,2020,33(10):5353-5367.
[21]刘芳军,李玥,武凌,等. 基于改进ResNet18的胡麻干旱胁迫识别与分类研究[J]. 江西农业大学学报,2023,45(6):1517-1527.
[22]XIA J, ZHANG W Y, ZHANG W X, et al. A cloud computing-based approach using the visible near-infrared spectrum to classify greenhouse tomato plants under water stress[J]. Computers and Electronics in Agriculture,2021,181:105966.
[23]KUO C E, TU Y K, FANG S L, et al. Early detection of drought stress in tomato from spectroscopic data:a novel convolutional neural network with feature selection[J]. Chemometrics and Intelligent Laboratory Systems,2023,239:104869.
[24]CHEN D J, NEUMANN K, FRIEDEL S, et al. Dissecting the phenotypic components of crop plant growth and drought responses based on high-throughput image analysis[J]. The Plant Cell,2014,26(12):4636.
[25]KAUTZ B, NOGA G, HUNSCHE M. Sensing drought-and salinity-imposed stresses on tomato leaves by means of fluorescence techniques[J]. Plant Growth Regulation,2014,73(3):279-288.
[26]MISHRA K B, IANNACONE R, PETROZZA A, et al. Engineered drought tolerance in tomato plants is reflected in chlorophyll fluorescence emission[J]. Plant Science,2012,182:79-86.
[27]WANG Y, AN Y, YU J, et al. Different responses of photosystem Ⅱ and antioxidants to drought stress in two contrasting populations of sour jujube from the Loess Plateau,China[J]. Ecological Research,2016,31(6):761-775.
[28]GUO Y Y, YU H Y, KONG D S, et al. Effects of drought stress on growth and chlorophyll fluorescence of Lycium ruthenicum murr. seedlings[J]. Photosynthetica,2016,54(4):524-531.
[29]LOTFI R, PESSARAKLI M, GHARAVI-KOUCHEBAGH P, et al. Physiological responses of brassica napus to fulvic acid under water stress:chlorophyll a fluorescence and antioxidant enzyme activity[J]. Crop Journal,2015,3(5):434-439.
[30]SHARMA D K, ANDERSEN S B, OTTOSEN C O, et al. Wheat cultivars selected for high Fv/Fm under heat stress maintain high photosynthesis,total chlorophyll,stomatal conductance,transpiration and dry matter[J]. Physiologia Plantarum,2015,153(2):284-298.
[31]HUMPLIK J F, LAZAR D, FURST T, et al. Automated integrative high-throughput phenotyping of plant shoots:a case study of the cold-tolerance of pea (Pisum sativum L.)[J]. Plant Methods,2015,11:1-11.
[32]BRESSON J, VASSEUR F, DAUZAT M, et al. Interact to survive:phyllobacterium brassicacearum improves Arabidopsis tolerance to severe water deficit and growth recovery[J]. PLoS One,2014,9(9):e107607.
[33]梁欢,韦宝,陈静,等. 基于叶绿素荧光参数的紫花苜蓿种质苗期抗旱性评价[J]. 草地学报,2020,28(1):45-55.
[34]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.
[35]李莉,万正煌,陈宏伟,等. 不同群体密度对绿豆农艺性状和产量的影响[J]. 农业科学与技术,2010,11(7):62-65.
[36]王小英,王孟,王斌,等. 不同氮肥用量对绿豆主要农艺性状及产量的影响[J]. 中国农学通报,2020,36(17):95-98.
[37]冯蕊,周琪,吴令上,等. PEG6000模拟干旱胁迫对铁皮石斛幼苗生理和叶绿素荧光特性的影响[J]. 浙江农林大学学报,2024,41(1):132-144.
[38]浦婵,董文渊,张孟楠,等. 聚乙二醇6000模拟干旱胁迫及复水对黄竹幼苗抗性生理特性的影响[J]. 东北林业大学学报,2018,46(8):31-37.
[39]姚松林. 基于卷积神经网络的荔枝病虫害图像识别技术研究[J]. 电脑编程技巧与维护,2023,12:133-135.
[40]李静,徐其江. 干旱胁迫对绿豆农艺性状及产量的影响研究[J]. 新疆农垦科技,2017,40(7):8-11.
[41]冯晓硕,沈樾,王冬琦. 基于图像的数据增强方法发展现状综述[J]. 计算机科学与应用,2021,11(2):13.
[42]温艳兰,陈友鹏,王克强,等. 基于机器视觉的病虫害检测综述[J]. 中国粮油学报,2022,37(10):271-279.

相似文献/References:

[1]佚名 佚名 佚名.三才期刊采编系统文章正在整理中…[J].江苏农业学报,2005,(01):5.
　XIE Xin,ZHAO Zhong.三才期刊采编系统文章正在整理中…[J].,2005,(01):5.
[2]佚名 佚名 佚名.三才期刊采编系统文章正在整理中…[J].江苏农业学报,2006,(01):5.
　XIE Xin,ZHAO Zhong.三才期刊采编系统文章正在整理中…[J].,2006,(01):5.
[3]熊洁,邹晓芬,邹小云,等.干旱胁迫对不同基因型油菜农艺性状和产量的影响[J].江苏农业学报,2015,(03):494.[doi:10.3969/j.issn.1000-4440.2015.03.005]
　XIONG Jie,ZOU Xiao-fen,ZOU Xiao-yun,et al.Effects of drought stress on agronomic traits and yield of different rapeseed genotypes[J].,2015,(01):494.[doi:10.3969/j.issn.1000-4440.2015.03.005]
[4]肇莹,杨镇,杨涛,等.植物内生菌醇提取物对草坪草抗旱性的影响[J].江苏农业学报,2015,(01):39.[doi:10.3969/j.issn.1000-4440.2015.01.006]
　ZHAO Ying,YANG Zhen,YANG Tao,et al.Influence of plant endophyte extract on turfgrass drought tolerance[J].,2015,(01):39.[doi:10.3969/j.issn.1000-4440.2015.01.006]
[5]孟力力,张俊,闻婧.干旱胁迫对彩叶草光合特性及叶片超微结构的影响[J].江苏农业学报,2015,(01):180.[doi:10.3969/j.issn.1000-4440.2015.01.028]
　MENG Li-li,ZHANG Jun,WEN Jing.Changes of photosynthetic characteristics of Coleus blumei and mesophyll cell ultrastructure in response to drought stress[J].,2015,(01):180.[doi:10.3969/j.issn.1000-4440.2015.01.028]
[6]张丽丽,徐碧玉,刘菊华,等.MaASR1基因通过乙烯途径提高拟南芥抗旱性的作用机制[J].江苏农业学报,2018,(03):511.[doi:doi:10.3969/j.issn.1000-4440.2018.03.005]
　ZHANG Li-li,XU Bi-yu,LIU Ju-hua,et al.The regulation mechanism of MaASR1 gene for improving the drought resistance of Arabidopsis by ethylene pathway[J].,2018,(01):511.[doi:doi:10.3969/j.issn.1000-4440.2018.03.005]
[7]麻云霞,李钢铁,张宏武,等.外源硅对酸枣生长和生理生化特征的影响[J].江苏农业学报,2018,(05):1113.[doi:doi:10.3969/j.issn.1000-4440.2018.05.021]
　MA Yun-xia,LI Gang-tie,ZHANG Hong-wu,et al.Effects of exogenous silicon on growth， physiological and biochemical characteristics of zizyphus jujube plant[J].,2018,(01):1113.[doi:doi:10.3969/j.issn.1000-4440.2018.05.021]
[8]葛道阔,曹宏鑫,杨余旺,等.干旱胁迫下油菜栽培模拟优化决策系统（Rape-CSODS）的订正及其检验[J].江苏农业学报,2019,(01):56.[doi:doi:10.3969/j.issn.1000-4440.2019.01.008]
　GE Dao-kuo,CAO Hong-xin,YANG Yu-wang,et al.Modification and verification of Rape-CSODS under drought stress[J].,2019,(01):56.[doi:doi:10.3969/j.issn.1000-4440.2019.01.008]
[9]陈丽,焦健,朱绍丹,等.油橄榄对牧草间作与干旱胁迫交互作用的根系生理响应[J].江苏农业学报,2019,(06):1434.[doi:doi:10.3969/j.issn.1000-4440.2019.06.024]
　CHEN Li,JIAO Jian,ZHU Shao-dan,et al.Root physiological response of olive to the interaction of pasture intercropping and drought stress[J].,2019,(01):1434.[doi:doi:10.3969/j.issn.1000-4440.2019.06.024]
[10]陈丽,焦健,朱绍丹,等.牧草间作对干旱胁迫下油橄榄根系形态特征的影响[J].江苏农业学报,2020,(01):39.[doi:doi:10.3969/j.issn.1000-4440.2020.01.006]
　CHEN Li,JIAO Jian,ZHU Shao-dan,et al.Effects of pasture intercropping on root morphological characteristics of olive under drought stress[J].,2020,(01):39.[doi:doi:10.3969/j.issn.1000-4440.2020.01.006]
[11]周旭旭,刘金洋,陈新,等.绿豆Alfin1-like基因家族的鉴定与干旱胁迫下的表达分析[J].江苏农业学报,2022,38(05):1179.[doi:doi:10.3969/j.issn.1000-4440.2022.05.004]
　ZHOU Xu-xu,LIU Jin-yang,CHEN Xin,et al.Identification of Alfin1-like gene family in Vigna radiata (L.) Wilczek and its expression analysis under drought stress[J].,2022,38(01):1179.[doi:doi:10.3969/j.issn.1000-4440.2022.05.004]

备注/Memo

备注/Memo:
收稿日期:2024-04-07基金项目:国家食用豆产业技术体系岗位科学家项目(CARS-09-G13)；江苏省种业揭榜挂帅项目[JBGS(2021)004]；江苏省研究生实践创新计划项目(SJCX24_2146)作者简介:蒋东山(1998-),男,江苏南京人,硕士,研究方向为深度学习、计算机视觉。(E-mail)dongshanJiang2022@163.com。刘金洋为共同第一作者。通讯作者:高尚兵,(E-mail)11060036@hyit.edu.cn；袁星星,(E-mail)20090049@jaas.ac.cn
更新日期/Last Update: 2025-02-28