[1]张善文,张晴晴,李萍.基于改进深度卷积神经网络的苹果病害识别[J].林业工程学报,2019,4(04):107-112.[doi:10.13360/j.issn.2096-1359.2019.04.016]
 ZHANG Shanwen,ZHANG Qingqing,LI Ping.Apple disease identification based on improved deep convolutional neural network[J].Journal of Forestry Engineering,2019,4(04):107-112.[doi:10.13360/j.issn.2096-1359.2019.04.016]
点击复制

基于改进深度卷积神经网络的苹果病害识别()
分享到:

《林业工程学报》[ISSN:1001-8081/CN:32-1160/S]

卷:
4
期数:
2019年04期
页码:
107-112
栏目:
装备与信息化
出版日期:
2019-07-09

文章信息/Info

Title:
Apple disease identification based on improved deep convolutional neural network
文章编号:
2096-1359(2019)04-0107-06
作者:
张善文张晴晴李萍
郑州西亚斯学院,郑州 451150
Author(s):
ZHANG Shanwen ZHANG Qingqing LI Ping
Zhengzhou SIAS University, Zhengzhou 451150, China
关键词:
苹果病害 图像识别 深度卷积神经网络 全局平均池化
Keywords:
apple disease image recognition deep convolutional neural networks(DCNNs) global average pooling
分类号:
TP391.4; S431.9
DOI:
10.13360/j.issn.2096-1359.2019.04.016
文献标志码:
A
摘要:
传统的深度卷积神经网络(DCNNs)使用3个全连接层将经过多次卷积层和池化层后提取到的特征图映射并连接为一个特征向量,然后利用Softmax分类器进行分类。该模型容易出现过拟合问题,而且由于在全连接层中参数太多,导致训练时间增加和泛化能力下降。针对传统的DCNNs模型在图像识别中出现的问题,提出一种改进的DCNNs模型,并应用于苹果叶部病害识别中。相比传统的DCNNs算法,改进的DCNNs利用一个全局平均池化层替代全连接层,并利用改进的Softmax分类器进行病害类别识别。在苹果病害叶片图像数据库上的实验结果表明,该模型能够克服过拟合问题,提高病害的识别率,大幅度降低模型的训练和识别时间。
Abstract:
Deep convolutional neural networks(DCNNs)have achieved impressive performance in the field of image classification.In the classical DCNNs, there are often two or three fully connected layers to be used to map the learned feature maps and concatenate them as a one-dimensional feature vector, and then are input into the Softmax classifier to classify the images.The drawbacks of these DCNNs models are the over-fitting phenomenon, long time to train their parameters and weak generalization ability due to too many parameters in the fully connected layers, especially in the last fully connected layer.Because there are too many parameters in the model and limited training samples in practice, the trained model is easy to produce over-fitting.Over-fitting problem can be solved by Dropout.Because the weight update obtained by Dropout is no longer dependent on the joint action of hidden nodes with fixed relations, preventing certain features from having effects only under other specific features, therefore the network can learn more robust features.Although dropout is effective to handle the over-fitting, one of the disadvantages is that the cost function is no longer clearly defined.Because in iterations, some points are randomly removed, it is hard to check the gradient descent performance.In order to improve the performance of DCNNs-based image recognition method, an improved DCNNs(IDCNNs)model is proposed and applied to apple disease identification.Different from the classical DCNNs, the proposed model employs one global average pooling layer instead of all the full connection layers and utilizes the modified Softmax classifier replacing the Softmax classifier of DCNNs.IDCNNs have two advantages.One is that there are few parameters needing to be optimized in the global pooling layer, thus over-fitting is naturally avoided at this layer.The other is that the global pooling sums up the spatial information, hence it is more robust to spatial translation of the input images.Moreover, IDCNNs is more consistent with the working structure of DCNNs.Each feature map is correlated with the category output, rather than the unit of the feature map is directly correlated with the category output.Experimental results on an actual apple diseased leaf image database show that the proposed method can address the overfitting problem, improve the recognition rate of disease recognition and greatly reduce the training and recognition time of the model.

参考文献/References:

[1] IQBAL M, NAEEM M, AZIZ U, et al.An overview of cotton leaf curl virus disease, persistant challenge for cotton production[J].Bulgarian Journal of Agricultural Science, 2014, 20(2014): 405-415.
[2] FANG Y, RAMASAMY R P.Current and prospective methods for plant disease detection[J].Biosensors, 2015, 5(3): 537-561.DOI:10.3390/bios5030537.
[3] MARTINELLI F, SCALENGHE R, DAVINO S, et al.Advanced methods of plant disease detection.a review[J].Agronomy for Sustainable Development, 2015, 35(1): 1-25.DOI:10.1007/s13593-014-0246-1.
[4] PATIL J K, KUMAR R.Color feature extraction of tomato leaf diseases[J].International Journal of Engineering Trends and Technology, 2011, 2(2):73-75.
[5] 毕傲睿.苹果叶子病害图像识别系统的设计与实现[D].西安: 西安建筑科技大学, 2014.
BI A R.Design and implementation of apple-leaf disease image recognition system[D].Xi'an: Xi'an University of Architecture and Technology, 2014.
[6] 屈赟, 陶晡, 王政嘉, 等.基于Android的苹果叶部病害识别系统设计[J].河北农业大学学报, 2015, 38(6): 102-106.DOI:10.13320/j.cnki.jauh.2015.0144.
QU Y, TAO B, WANG Z J, et al.Design of apple leaf disease recognition system based on Android[J].Journal of Agricultural University of Hebei, 2015, 38(6): 102-106.
[7] DUBEY S R, JALAL A S.Apple disease classification using color, texture and shape features from images[J].Signal, Image and Video Processing, 2016, 10(5): 819-826.DOI:10.1007/s11760-015-0821-1.
[8] 李超, 彭进业, 张善文.基于特征融合与局部判别映射的苹果叶部病害识别方法[J].广东农业科学, 2016, 43(10): 134-139.DOI:10.16768/j.issn.1004-874X.2016.10.024.
LI C, PENG J Y, ZHANG S W.Apple leaf disease recognition based on feature fusion and local discriminant projection[J].Guangdong Agricultural Sciences, 2016, 43(10): 134-139.
[9] 师韵, 黄文准, 张善文.基于二维子空间的苹果病害识别方法[J].计算机工程与应用, 2017, 53(22): 180-184.DOI:10.3778/j.issn.1002-8331.1605-0073.
SHI Y, HUANG W Z, ZHANG S W.Apple disease recognition based on two-dimensionality subspace learning[J].Computer Engineering and Applications, 2017, 53(22): 180-184.
[10] 党荣辉.基于图像分析的陕西苹果叶片病害识别系统设计与实现[D].杨凌:西北农林科技大学, 2017.
DANG R H.Design and implementation of shaanxi apple leaf disease recognition system based on image analysis [D].Yangling: Northwest University of Agriculture and Forestry, 2017.
[11] 孙素云.基于图像处理和支持向量机的苹果树叶部病害的分类研究[D].西安: 西安科技大学, 2017.
SUN S Y.Classification of apple leaf diseases based on image processing and support vector machine [D].Xi'an: Xi'an University of Science and Technology, 2017.
[12] SCHMIDHUBER J.Deep learning in neural networks: an overview[J].Neural Networks, 2015, 61:85-117.DOI:10.1016/j.neunet.2014.09.003
[13] 孙俊, 谭文军, 毛罕平, 等.基于改进卷积神经网络的多种植物叶片病害识别[J].农业工程学报, 2017, 33(19): 209-215.DOI:10.11975/j.issn.1002-6819.2017.19.027.
SUN J, TAN W J, MAO H P, et al.Recognition of multiple plant leaf diseases based on improved convolutional neural network[J].Transactions of the Chinese Society of Agricultural Engineering, 2017, 33(19): 209-215.
[14] KRIZHEVSKY A, SUTSKEVER I, HINTON G E.ImageNet classification with deep convolutional neural networks[C]//Proceedings of the 25th International Conference on Neural Information Processing Systems-Volume 1, 2012: 1097-1105.
[15] SLADOJEVIC S, ARSENOVIC M, ANDERLA A, et al.Deep neural networks based recognition of plant diseases by leaf image classification[J].Computational Intelligence and Neuroscience, 2016, 2016: 1-11.DOI:10.1155/2016/3289801.
[16] AMARA J, BOUAZIZ B, ALGERGAWY A A.A deep learning-based approach for banana leaf diseases classification[C]//Datenbanksysteme Für Business, Technologie Und Web, 2017:79-88.
[17] ANANDHAKRISHNAN M G, JOEL H, ANNETTE J, et al.Plant leaf disease detection using deep learning and convolutional neural network [J].International Journal of Engineering Science and Computing, 2017,7(3):5324-5328.
[18] FERENTINOS K P.Deep learning models for plant disease detection and diagnosis[J].Computers and Electronics in Agriculture, 2018, 145: 311-318.DOI:10.1016/j.compag.2018.01.009.
[19] LIU B, ZHANG Y, HE D J, et al.Identification of apple leaf diseases based on deep convolutional neural networks[J].Symmetry, 2017, 10(1): 11.DOI:10.3390/sym10010011.
[20] PAWARA P, OKAFOR E, SCHOMAKER L, et al.Data augmentation for plant classification[C]//Blanc-Talon J, Penne R, Philips W, et al.Advanced Concepts for Intelligent Vision Systems.ACIVS 2017.Lecture Notes in Computer Science, vol 10617.Springer, Cham.DOI: 10.1007/978-3-319-70353-4_52.
[21] 冉鹏, 王灵, 李昕, 等.改进Softmax分类器的深度卷积神经网络及其在人脸识别中的应用[J].上海大学学报(自然科学版), 2018, 24(3): 352-366.DOI:10.12066/j.issn.1007-2861.1831.
RAN P, WANG L, LI X, et al.Improved softmax classifier for deep convolution neural networks and its application in face recognition[J].Journal of Shanghai University(Natural Science Edition), 2018, 24(3): 352-366.

相似文献/References:

[1]张习之,李立君*.基于改进卷积自编码机的油茶果图像识别研究[J].林业工程学报,2019,4(03):118.[doi:10.13360/j.issn.2096-1359.2019.03.018]
 ZHANG Xizhi,LI Lijun*.Research of image recognition of camellia oleifera fruit based on improved convolutional auto-encoder[J].Journal of Forestry Engineering,2019,4(04):118.[doi:10.13360/j.issn.2096-1359.2019.03.018]

备注/Memo

备注/Memo:
收稿日期:2018-12-30 修回日期:2019-02-22
基金项目:国家自然科学基金(61473273); 河南省教育厅科技攻关项目(182102210544,182102311094)。
作者简介:张善文,男,教授,研究方向为智能算法处理及其应用。E-mail: wjdw716@163.com
更新日期/Last Update: 2019-07-10