当前位置>主页 > 期刊在线 > 信息技术 >

信息技术21年9期

基于卷积神经网络的玉米病害识别
吴淑琦
(聊城大学 季羡林学院,山东 聊城 252000)

摘  要:在玉米生长初期,不能及时知道玉米所患病害从而无法及时医治,将导致玉米产量和质量下降。而人工分辨玉米病害耗费大量人力和时间,判断准确率也不高。因此文章提出了基于卷积神经网络的玉米病害识别模型,模型主要有 12 个网络层,其中包含输入层、4 个卷积层、4 个池化层、2 个全连接层和输出层。通过调整参数和模型优化等操作,最终分类准确率在 95% 左右。模型具有一定的实际意义,可为玉米病害防治提供理论依据。


关键词:玉米;卷积神经网络;模型;病害



DOI:10.19850/j.cnki.2096-4706.2021.09.002


中图分类号:TP183;TP391.4                          文献标识码:A                                文章编号:2096-4706(2021)09-0006-04


Corn Disease Identification Based on Convolution Neural Network

WU Shuqi

(Ji Xianlin Honors School of Liaocheng University,Liaocheng 252000,China)

Abstract:In the early growth of corn,we can not know the diseases of corn in time and thus cannot be cured in time,which leads to a decline in the yield and quality of corn. However,it takes a lot of manpower and time to distinguish corn diseases manually,and the accuracy of judgment is not high. Therefore,a corn disease recognition model based on convolutional neural network is proposed in this paper. The model mainly has 12 network layers,including input layer,4 convolutional layers,4 pooling layers,2 fully connected layers and output layer. By adjusting the parameters and the optimization model,the final classification accuracy is about 95%. The model has certain practical significance and can provide a theoretical basis for corn disease control.

Keywords:corn;convolutional neural network;model;disease


参考文献:

[1] 章楷,李根蟠 . 玉米在我国粮食作物中地位的变化——兼 论我国玉米生产的发展和人口增长的关系 [J]. 农业考古,1983(2): 94-99.

[2] 王晓鸣,晋齐鸣,石洁,等 . 玉米病害发生现状与推广 品种抗性对未来病害发展的影响 [J]. 植物病理学报,2006(1): 1-11.

[3] 刘翱宇,吴云志,朱小宁,等 . 基于深度残差网络的玉米 病害识别 [J]. 江苏农业学报,2021,37(1):67-74.

[4] 许景辉,邵明烨,王一琛,等 . 基于迁移学习的卷积神经 网络玉米病害图像识别 [J]. 农业机械学报,2020,51(2):230- 236+253.

[5] 顾博,邓蕾蕾,李巍,等 . 基于 GrabCut 算法的玉米病害 图像识别方法研究 [J]. 中国农机化学报,2019,40(11):143- 149.

[6] O’SHIEA K,NASH R. An Introduction to Convolutional Neural Networks [J/OL].arXiv:1511.08458 [cs.NE].(2015-11-26). https://arxiv.org/abs/1511.08458.


作者简介:吴淑琦(2001—),女,汉族,山东阳谷人,本科 在读,研究方向:模式识别。