Identification of plant disease images via a squeeze-and-excitation MobileNet model and twice transfer learning

被引:55
作者
Chen, Junde [1 ]
Zhang, Defu [1 ]
Suzauddola, Md [1 ]
Nanehkaran, Yaser Ahangari [1 ]
Sun, Yuandong [2 ]
机构
[1] Xiamen Univ, Sch Informat, Xiamen 361005, Peoples R China
[2] Spira Inc, Los Angeles, CA USA
关键词
Image enhancement;
D O I
10.1049/ipr2.12090
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Crop diseases have a devastating effect on agricultural production, and serious diseases can lead to harvest failure entirely. Recent developments in deep learning have greatly improved the accuracy of image identification. In this study, we investigated the transfer learning of deep convolutional neural networks and modified the network structure to improve the learning capability of plant lesion characteristics. The MobileNet with squeeze-and-excitation (SE) block was selected in our approach. Integrating the merits of both, the pre-trained MobileNet and SE block were fused to form a new network, which we termed the SE-MobileNet, and was used to identify the plant diseases. In particular, the transfer learning was performed twice to obtain the optimum model. The first phase trained the model for the extended layers while the bottom convolution layers were frozen with the pre-trained weights on ImageNet; by loading the model trained in the first phase, the second phase retrained the model using the target dataset. The proposed procedure provides a significant increase in efficiency relative to other state-of-the-art methods. It reaches an average accuracy of 99.78% in the public dataset with clear backdrops. Even under multiple classes and heterogeneous background conditions, the average accuracy realises 99.33% for identifying the rice disease types. The experimental findings show the feasibility and effectiveness of the proposed procedure.
引用
收藏
页码:1115 / 1127
页数:13
相关论文
共 37 条
[1]  
Amara Jihen, 2017, DAT BUS TECHN WEB BT
[2]   Solving Current Limitations of Deep Learning Based Approaches for Plant Disease Detection [J].
Arsenovic, Marko ;
Karanovic, Mirjana ;
Sladojevic, Srdjan ;
Anderla, Andras ;
Stefanovic, Darko .
SYMMETRY-BASEL, 2019, 11 (07)
[3]   Factors influencing the use of deep learning for plant disease recognition [J].
Barbedo, Jayme G. A. .
BIOSYSTEMS ENGINEERING, 2018, 172 :84-91
[4]   Deep Learning for Tomato Diseases: Classification and Symptoms Visualization [J].
Brahimi, Mohammed ;
Boukhalfa, Kamel ;
Moussaoui, Abdelouahab .
APPLIED ARTIFICIAL INTELLIGENCE, 2017, 31 (04) :299-315
[5]   Detection of rice plant diseases based on deep transfer learning [J].
Chen, Junde ;
Zhang, Defu ;
Nanehkaran, Yaser A. ;
Li, Dele .
JOURNAL OF THE SCIENCE OF FOOD AND AGRICULTURE, 2020, 100 (07) :3246-3256
[6]  
Durmus H, 2017, INT CONF AGRO-GEOINF, P46
[7]   Plant identification using deep neural networks via optimization of transfer learning parameters [J].
Ghazi, Mostafa Mehdipour ;
Yanikoglu, Berrin ;
Aptoula, Erchan .
NEUROCOMPUTING, 2017, 235 :228-235
[8]  
Glorot X., 2011, JMLR WORKSHOP C P, P315
[9]  
Goodfellow IJ, 2014, ADV NEUR IN, V27, P2672
[10]  
Howard A.G., 2017, MOBILENETS EFFICIENT