Scholarlink

Chrysanthemums family are the most important Economical plants due to its Ornamental, Decorative, and Medicinal Applications. Numerous environmental factors can negatively impact the growth and well-being of these plants. By utilizing cutting-edge deep learning models such as VGG16, ResNet50, MobileNetV2, and DenseNet121, a novel method for chrysanthemum leaf phenotyping and stress evaluation is proposed. A broad dataset consisting of images of chrysanthemum leaves is gathered for the study and is used along with data augmentation to model stressful situations. For stress comparison and evaluation, the enhanced images are subsequently loaded into the above-mentioned deep learning models. Grad-CAM (Gradient-weighted Class Activation Mapping) is used to get insights into how these models make decisions. Grad-CAM creates heat maps that highlight the leaf image regions that are essential for stress categorization, allowing researchers to understand and see the model’s predictions. Experimental results obtained for various deep learning models indicated that the VGG 16 model outperforms the others with 68% average stress for 0.5 stress threshold.

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

Analysis of Stress in Chrysanthemum Family Using Grad-CAM with CNN

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Dhanagopalan Kannan,
  • Balasundaram Revathi Alias Ponmozhi

摘要

Chrysanthemums family are the most important Economical plants due to its Ornamental, Decorative, and Medicinal Applications. Numerous environmental factors can negatively impact the growth and well-being of these plants. By utilizing cutting-edge deep learning models such as VGG16, ResNet50, MobileNetV2, and DenseNet121, a novel method for chrysanthemum leaf phenotyping and stress evaluation is proposed. A broad dataset consisting of images of chrysanthemum leaves is gathered for the study and is used along with data augmentation to model stressful situations. For stress comparison and evaluation, the enhanced images are subsequently loaded into the above-mentioned deep learning models. Grad-CAM (Gradient-weighted Class Activation Mapping) is used to get insights into how these models make decisions. Grad-CAM creates heat maps that highlight the leaf image regions that are essential for stress categorization, allowing researchers to understand and see the model’s predictions. Experimental results obtained for various deep learning models indicated that the VGG 16 model outperforms the others with 68% average stress for 0.5 stress threshold.