[1]梁凯博,孙立,汪禹治,等.基于超轻量化卷积神经网络的番茄病虫害诊断[J].江苏农业学报,2024,(03):438-449.[doi:doi:10.3969/j.issn.1000-4440.2024.03.006]
　LIANG Kai-bo,SUN Li,WANG Yu-zhi,et al.Diagnosis of tomato pests and diseases based on super lightweight convolutional neural network[J].,2024,(03):438-449.[doi:doi:10.3969/j.issn.1000-4440.2024.03.006]
点击复制

基于超轻量化卷积神经网络的番茄病虫害诊断()
分享到：

江苏农业学报[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2024年03期
页码:
438-449
栏目:
植物保护
出版日期:
2024-03-30

文章信息/Info

Title:
Diagnosis of tomato pests and diseases based on super lightweight convolutional neural network
作者:
梁凯博1; 2; 孙立1; 汪禹治2; 靳龙豪1; 燕雪倩3; 曾旺1
（1.北京物资学院信息学院，北京101149；2.首都经济贸易大学管理工程学院，北京100070；3.北京物资学院数据科学与统计学院，北京101149）
Author(s):
LIANG Kai-bo1; 2; SUN Li1; WANG Yu-zhi2; JIN Long-hao1; YAN Xue-qian3; ZENG Wang1
(1.School of Information, Beijing Wuzi University, Beijing 101149, China;2.School of Management Engineering, Capital University of Economics and Business, Beijing 100070, China;3.School of Data Science and Statistics, Beijing Wuzi University, Beijing 101149, China)
关键词:
图像识别; 番茄病虫害; 超轻量化卷积神经网络; 不平衡性
Keywords:
image recognition; tomato pests and diseases; super lightweight convolutional neural network; imbalance
分类号:
S641.2;TP391.41
DOI:
doi:10.3969/j.issn.1000-4440.2024.03.006
摘要:
针对番茄病虫害诊断中存在的传统卷积神经网络结构复杂、难以直接应用于便携终端，以及现有轻量化卷积神经网络特征提取能力弱、识别准确率低、难以满足实际需要等问题，本研究拟在原有轻量化卷积神经网络的基础上，定义超轻量化卷积神经网络，设计一种基于SqueezeNet网络改进的超轻量化卷积神经网络，将其用于番茄病虫害诊断任务中。首先，改进SqueezeNet网络中的Fire模块，生成2种适用于不同特征维度的Fire模块，并引入ECA（高效通道注意力）模块以提高模型的特征提取能力；其次，结合扩展型指数线性单元函数（SELU）和Mish函数，替代修正线性单元函数（ReLU）作为激活函数；再次，采用软池化（Softpool）替代原始的最大池化；最后，利用中心损失函数（Center loss）改进指数归一化损失函数（Softmax loss），提高对近似病虫害的识别准确率。本研究选择了8种害虫和9种病害，对害虫、病害、病虫害3类数据集进行数据增强，并探讨了数据的小样本性、不平衡性对模型性能的影响。结果表明，本研究提出的模型具有超轻量化的特点，对害虫、病害、病虫害的识别准确率最高分别可达98.83%、98.14%和97.71%，能够很好地满足番茄病虫害诊断需求。
Abstract:
In the diagnosis of tomato diseases and pests, traditional convolutional neural network structures are complex and hard to be directly applied to portable terminals. Besides, existing lightweight convolutional neural networks exhibit weak feature extraction capabilities, low recognition accuracy, and are inadequate for practical applications. Aiming at the above problems, we intended to define a super lightweight convolutional neural network based on existing lightweight convolutional neural network, and to design an ultra-lightweight convolutional neural network by improving the SqueezeNet network for tomato disease and pest diagnosis tasks. Firstly, we enhanced the Fire module in the SqueezeNet network, generated two Fire modules suitable for different feature dimensions. We introduced efficient channel attention (ECA) module to improve feature extraction capabilities of the model. Secondly, we incorporated scaled exponential linear unit (SELU) and Mish to replace rectified linear unit (ReLU) as activation function. Next, we employed Softpool instead of the original max pooling. Finally, we enhanced the exponential normalized loss (Softmax loss) by using Center loss function to improve the recognition accuracy of approximate diseases and pests. In this experiment, we selected eight types of pests and nine types of diseases to perform data augmentation on three datasets (pests, diseases, diseases and pests), and investigated the impact of small sample and data imbalance on model performance. Experimental results demonstrated that the network proposed in this study had super lightweight characteristics. The recognition accuracies for pests, diseases, and diseases and pests could reach up to 98.83%, 98.14% and 97.71%, respectively, which met the requirements for diagnosis effectively.

参考文献/References:

[1]PIERCE F J, NOWAK P. Aspects of precision agriculture[J]. Advances in Agronomy,1999,67:1-85.
[2]AGARWAL M, GUPTA S K, BISWAS K K. Development of efficient CNN model for tomato crop disease identification[J]. Sustainable Computing:Informatics and Systems,2020,28:100407.
[3]REN F, LIU W, WU G. Feature reuse residual networks for insect pest recognition[J]. IEEE Access,2019,7:122758-122768.
[4]彭红星,何慧君,高宗梅,等. 基于改进ShuffleNetV2模型的荔枝病虫害识别方法[J]. 农业机械学报,2022,53(12):290-300.
[5]曹乐平. 基于机器视觉的植物病虫害实时识别方法[J]. 中国农学通报,2015,31(20):244-249.
[6]ALVES A N, SOUZA W S R, BORGES D L. Cotton pests classification in field-based images using deep residual networks[J]. Computers and Electronics in Agriculture,2020,174:105488.
[7]BOLLIS E, MAIA H, PEDRINI H, et al. Weakly supervised attention-based models using activation maps for citrus mite and insect pest classification[J]. Computers and Electronics in Agriculture, 2022, 195: 106839.
[8]HUANG M L, CHUANG T C, LIAO Y C. Application of transfer learning and image augmentation technology for tomato pest identification[J]. Sustainable Computing:Informatics and Systems,2022,33:100646.
[9]周维,牛永真,王亚炜,等. 基于改进的YOLOv4-GhostNet水稻病虫害识别方法[J]. 江苏农业学报,2022,38(3):685-695.
[10]温长吉,王启锐,陈洪锐,等. 面向大规模多类别的病虫害识别模型[J]. 农业工程学报,2022,38(8):169-177.
[11]罗昱晟,冉力争,张小荣,等. 基于大数据与图像识别的柑橘病虫害防治系统的设计与实现[J]. 智能计算机与应用,2020,10(12):170-173.
[12]骆润玫,殷惠莉,刘伟康，等. 基于YOLOv5-C的广佛手病虫害识别[J]. 华南农业大学学报,2023,44(1):151-160.
[13]杨英茹,吴华瑞,张燕,等. 基于复杂环境的番茄叶部图像病虫害识别[J]. 中国农机化学报,2021,42(9):177-186.
[14]IANDOLA F N, HAN S, MOSKEWICZ M W, et al. SqueezeNet:AlexNet-level accuracy with 50x fewer parameters and <0.5 MB model size[J]. arXiv,2016. DOI:10.48550/arXiv.1602.07360.
[15]ERTAM F, YAKUT O F, TUNCER T. Pattern lock screen detection method based on lightweight deep feature extraction[J]. Neural Computing and Applications,2023,35(2):1549-1567.
[16]ZHONG H, LYU Y, YUAN R, et al. Bearing fault diagnosis using transfer learning and self-attention ensemble lightweight convolutional neural network[J]. Neurocomputing,2022,501:765-777.
[17]ZHONG Y, HAO J, LIU Q, et al. Novel diagnosis method for GIS mechanical defects based on an improved lightweight CNN model with load adaptive matching[J]. IEEE Transactions on Industrial Informatics,2023,19(11):11041-11051.
[18]MINU M, CANESSANE R, RAMESH S. Optimal squeeze net with deep neural network-based arial image classification model in unmanned aerial vehicles[J]. Traitement Du Signal,2022, 39(1):275-281.
[19]PANG Y, ZHANG Y, WANG Y, et al. SOCNet: a lightweight and fine-grained object recognition network for satellite on-orbit computing[J]. IEEE Transactions on Geoscience and Remote Sensing,2022,60.DOI:10.1109/TGRS.2022.3216215.
[20]LI Y, NIE J, CHAO X. Do we really need deep CNN for plant diseases identification?[J]. Computers and Electronics in Agriculture, 2020,178:105803.
[21]HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]. Salt Lake City,UT,USA:IEEE,2018.
[22]WANG Q, WU B, ZHU P, et al. ECA-Net:efficient channel attention for deep convolutional neural networks[C]. Seattle, WA, USA:IEEE,2020.
[23]WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]. Munich,Germany:Springer,2018.
[24]KLAMBAUER G, UNTERTHINER T, MAYR A, et al. Self-normalizing neural networks[J]. 2017.DOI:10.48550/arXiv.1706.02515.
[25]HENDRYCKS D, GIMPEL K. Gaussian error linear units (gelus) [J]. arXiv,2016.DOI:10.48550/arXiv.1606.08415.
[26]RAMACHANDRAM P, ZOPH B, LE Q V. Searching for activation functions[J]. arXiv,2017.DOI:10.48550/arXiv.1710.05941.
[27]MISRA D. Mish: a self regularized non-monotonic neural activation function[J]. arXiv,2019.DOI:10.48550/arXiv.1908.08681.
[28]STERGIOU A, POPPE R, KALLIATAKIS G. Refining activation downsampling with SoftPool[C]. Montreal, Canada: IEEE,2021.
[29]ZEILER M D, FERGUS R. Visualizing and understanding convolutional networks[C]. Zurich,Switzerland:Springer,2014.
[30]李超,李锋,黄炜嘉. 基于并联卷积神经网络的水果品种识别[J]. 浙江农业学报,2022,34(11):2533-2541.
[31]WEN Y, ZHANG K, LI Z, et al. A discriminative feature learning approach for deep face recognition[C]. Amsterdam, Netherland:Springer,2016.
[32]HUANG M, CHIN T. A database of eight commin tomato pest images[DB]. 2020.DOI:10.17632/s62zm6djd2.1.
[33]HUGHES D, SALATH’E M. An open access repository of images on plant health to enable the development of mobile disease diagnostics[J]. arXiv,2015.DOI:10.48550/arXiv.1511.08060.
[34]蒋清健,姚勇,付志军,等. 基于改进卷积神经网络算法的番茄叶片病害识别[J]. 江苏农业科学,2022,50(20):29-34.
[35]徐姗姗,吕净妍,陈芳媛. 深度卷积神经网络的遥感植被检测方法[J]. 南京林业大学学报(自然科学版),2022,46(4):185-193.
[36]牛学德,高丙朋,南新元,等. 基于改进DenseNet卷积神经网络的番茄叶片病害检测[J]. 江苏农业学报,2022,38(1):129-134.
[37]棘玉,尹显明,严恩萍,等. 基于相机拍照的油茶果形状特征提取研究[J]. 南京林业大学学报(自然科学版),2022,46(2):63-70.

相似文献/References:

[1]孙云云,江朝晖,董伟,等.基于卷积神经网络和小样本的茶树病害图像识别[J].江苏农业学报,2019,(01):48.[doi:doi:10.3969/j.issn.1000-4440.2019.01.007]
　SUN Yun-yun,JIANG Zhao-hui,DONG Wei,et al.Image recognition of tea plant disease based on convolutional neural network and small samples[J].,2019,(03):48.[doi:doi:10.3969/j.issn.1000-4440.2019.01.007]
[2]邱洪涛,孙裴,侯金波,等.基于Caffe的猪肉新鲜度分级的设计与实现[J].江苏农业学报,2019,(02):461.[doi:doi:10.3969/j.issn.1000-4440.2019.02.029]
　QIU Hong-tao,SUN Pei,HOU Jin-bo,et al.Design and implementation of pork freshness grading based on Caffe[J].,2019,(03):461.[doi:doi:10.3969/j.issn.1000-4440.2019.02.029]
[3]牛学德,高丙朋,南新元,等.基于改进DenseNet卷积神经网络的番茄叶片病害检测[J].江苏农业学报,2022,38(01):129.[doi:doi:10.3969/j.issn.1000-4440.2022.01.015]
　NIU Xue-de,GAO Bing-peng,NAN Xin-yuan,et al.Detection of tomato leaf disease based on improved DenseNet convolutional neural network[J].,2022,38(03):129.[doi:doi:10.3969/j.issn.1000-4440.2022.01.015]

备注/Memo

备注/Memo:
收稿日期：2023-03-31基金项目：广东省重点领域研发计划项目（2019B020214002）；北京市社会科学基金项目（20GLB026）；国家自然科学基金项目（71771028）；首都经济贸易大学研究生科技创新项目（2023KJCX062）作者简介：梁凯博（1997-），男，河北秦皇岛人，博士研究生，研究方向为深度学习、人工智能。（E-mail）liangkaibo2955@gmail.com通讯作者：孙立，（E-mail）slsally@163.com
更新日期/Last Update: 2024-05-20