Scholarlink

<p>Detection of plant leaf diseases is a necessary measure to be followed in the field of agriculture to increase the productivity of crops thereby increasing food productivity. Thus, this study proposes a new deep learning paradigm known as the Gradient Weighted DenseNet-201 for plant leaf disease classification. In this modal, DenseNet-201 is integrated with Gradient Weighted Class Activation Mapping which enhances the feature extraction performance and achieves a higher classification rate. Further, image texture features such as the Gray Level Co-occurrence Matrix are used to represent spatial relationships between the picture elements and facilitate accurate identification of affected regions. The proposed model is evaluated using four benchmark datasets: Corn or Maize Leaf Disease Dataset, Banana Leaf Spot Diseases Dataset, Rose Leaf Disease Dataset, and Rice Leaf Diseases Dataset. The main preprocessing steps, like grayscale conversion, smoothing, Otsu threshold segmentation, and morphology are applied to have better and highest quality input data. All of these experimental outcomes show that the proposed GradWDN-201 model is more accurate than the conventional models experimented with an accuracy of 0.9896, precision of 0.9757, BF score of 0.9786, and ROC-AUC of 0.9912. These superior values demonstrate the effectiveness of the proposed model in the classification of plant leaf disease.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Advancing agricultural sustainability with gradient weighted Densenet-201 model for accurate detection of plant leaf diseases

  • 期刊论文

  • 发表日期：

    2025年5月31日

查看全文
  • Senthil Pandi S,
  • M. Pounambal,
  • Jagadeesh G,
  • J. Vellingiri,
  • ArivuSelvan K

摘要

Detection of plant leaf diseases is a necessary measure to be followed in the field of agriculture to increase the productivity of crops thereby increasing food productivity. Thus, this study proposes a new deep learning paradigm known as the Gradient Weighted DenseNet-201 for plant leaf disease classification. In this modal, DenseNet-201 is integrated with Gradient Weighted Class Activation Mapping which enhances the feature extraction performance and achieves a higher classification rate. Further, image texture features such as the Gray Level Co-occurrence Matrix are used to represent spatial relationships between the picture elements and facilitate accurate identification of affected regions. The proposed model is evaluated using four benchmark datasets: Corn or Maize Leaf Disease Dataset, Banana Leaf Spot Diseases Dataset, Rose Leaf Disease Dataset, and Rice Leaf Diseases Dataset. The main preprocessing steps, like grayscale conversion, smoothing, Otsu threshold segmentation, and morphology are applied to have better and highest quality input data. All of these experimental outcomes show that the proposed GradWDN-201 model is more accurate than the conventional models experimented with an accuracy of 0.9896, precision of 0.9757, BF score of 0.9786, and ROC-AUC of 0.9912. These superior values demonstrate the effectiveness of the proposed model in the classification of plant leaf disease.