计算机工程与应用 ›› 2020, Vol. 56 ›› Issue (23): 124-129.DOI: 10.3778/j.issn.1002-8331.1909-0083

• 模式识别与人工智能 • 上一篇    下一篇

融合通道信息注意力网络的叶片病害识别

侯金秀,李然,邓红霞,李海芳   

  1. 太原理工大学 信息与计算机学院,太原 030600
  • 出版日期:2020-12-01 发布日期:2020-11-30

Leaf Disease Identification of Fusion Channel Information Attention Network

HOU Jinxiu, LI Ran, DENG Hongxia, LI Haifang   

  1. College of Information and Computer, Taiyuan University of Technology, Taiyuan 030600, China
  • Online:2020-12-01 Published:2020-11-30

摘要:

针对植物叶片病害存在的种类驳杂以及如何提取有效特征的问题,提出一种融合通道信息注意力网络模型对多种植物叶片病害进行识别。构建残差结构为主的基础网络用于特征提取,再将特征通过注意力网络融合多个通道信息对病害特征进行重标定;在交叉熵函数中添加约束信息以加快模型收敛速度;在4种不同植物16类病害的数据集上对该模型进行实验,结果表明,基础网络模型识别准确率为83.13%,而融合通道信息网络后准确率提高4.64个百分点;融合后的模型与其他模型相比在识别准确率方面提高9.72个百分点且模型复杂度约为对比实验中最优模型复杂度的1/2。

关键词: 叶片病害识别, 通道信息注意力网络, 模型融合, 卷积神经网络

Abstract:

In this paper, an attention network model with channel information is proposed to identify various plant leaf diseases. Firstly, a residual structure based basic network is built for feature extraction, and then the feature is re-calibrated by integrating multiple channel information through the attention network. Then, the constraint information is added to the cross entropy function to accelerate the convergence of the model. Finally, the model is tested on the data sets of 16 diseases of four different plants. The results show that the recognition accuracy of the basic network model is 83.13%, while the accuracy increases by 4.64 percentage points after the channel information network is fused. Compared with other models, the fusion model improves the recognition accuracy by 9.72 percentage points and the model complexity is about 1/2 of that of the optimal model in the comparison experiment.

Key words: identification of leaf disease, channel information attention network, model fusion, Convolutional Neural Network(CNN)