基于改进MaxViT的辣椒病害识别分类方法
CSTR:
作者:
作者单位:

湖北工业大学机械工程学院/湖北工业大学现代制造质量工程湖北省重点实验室,武汉 430068

作者简介:

李西兴,E-mail:li_xi_xing@126.com

通讯作者:

吴锐,E-mail:wurui@hbut.edu.cn

中图分类号:

TP391.41;S436.418.1

基金项目:

国家自然科学基金项目(51805152);湖北工业大学绿色工业引领计划项目(XJ2021005001);湖北省自然科学基金项目(2022CFB445);湖北省重点研发计划项目(2021BAA203)


A method for identifying and classifying pepper diseases based on improved MaxViT
Author:
Affiliation:

College of Mechanical Engineering/Key Laboratory of Modern Manufacturing Quality Engineering,Hubei University of Technology,Wuhan 430068,China

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • | |
  • 文章评论
    摘要:

    为实现复杂环境下辣椒病害的精准识别和分类,设计了一种适用于辣椒病害识别分类的方法。以辣椒在生长过程中常见的6种病害为分类研究的对象,使用数据增强的方法扩充数据集,提出一种基于MaxViT改进的MaxViT-DF模型,将MaxViT模型中的普通卷积替换为可变形卷积,使模型在提取特征时能更贴近复杂环境下的识别目标;同时在MaxViT模型施加注意力时引入特征融合模块,提高模型的全局感知能力。结果显示,改进的MaxViT-DF模型识别分类准确率达到98.10%,对6种辣椒病害的分类精度均高于95%。与ResNet-34、EfficientNetv2和VGG-16等模型相比,改进模型在收敛速度和分类精度上具有明显优势。以上结果表明,MaxViT-DF模型能够对不同种类的辣椒常见病害进行有效的分类识别。

    Abstract:

    A method suitable for identifying and classifying pepper diseases was designed to achieve precise recognition and classification of pepper diseases in complex environments including background clutter or interference.Six common diseases in the growth process of peppers were used to expand the dataset with data augmentation methods.An improved MaxViT-DF model was proposed based on MaxViT.The ordinary convolution in the MaxViT model was replaced with deformable convolution to enable the model to extract features closer to the recognition target in complex environments.A feature fusion module was introduced when applying attention to the MaxViT model to improve the model’s global perception ability.The results showed that the improved MaxViT-DF model had an identification and classification accuracy of 98.10%,and the classification accuracy for six common pepper diseases was higher than 95%.The improved model had significant advantages in convergence speed and classification accuracy compared with models including ResNet-34,EfficientNetv2,and VGG-16.It is indicated that the MaxViT-DF model can effectively identify and classify common diseases in different types of peppers.

    表 2 不同模型评估标准对比Table 2 Comparison of evaluation standards for different models
    表 1 各病害训练集来源Table 1 Source of each disease training set
    表 3 不同模型在各病害中的评估标准对比Table 3 Comparison of evaluation criteria for different models in different diseases
    图1 辣椒常见6种病害Fig.1 Common diseases of pepper
    图2 数据增强结果Fig.2 Data enhancement results
    图3 MaxViT模型Fig.3 MaxViT model
    图4 普通卷积和可变形卷积对比Fig.4 Comparison between ordinary convolutions and deformable convolutions
    图5 可变形卷积结构Fig.5 Structure of deformable convolution
    图6 特征融合模块结构Fig.6 Structure of feature fusion module
    图7 MaxViT-DF模型结构Fig.7 MaxViT-DF model structure
    图8 多轴注意力模块结构Fig.8 Structure diagram of multi-axis attention module
    图9 各分类模型损失曲线(A)和准确率曲线(B)Fig.9 Loss curves (A) and accuracy curve (B) in each classification model
    图10 基于MaxViT-DF模型的辣椒病害分类结果Fig.10 Classification results of pepper diseases based on MaxViT-DF model
    图11 各模型分类结果的混淆矩阵Fig.11 Confusion matrix of classification results of each model
    参考文献
    相似文献
    引证文献
引用本文

李西兴,陈佳豪,吴锐,杨睿.基于改进MaxViT的辣椒病害识别分类方法[J].华中农业大学学报,2024,43(2):123-133

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2023-06-05
  • 在线发布日期: 2024-04-02
文章二维码