Scholarlink

<p>Plant pathogens pose a significant risk to the security of food supplies worldwide. Plant disease identification is still difficult and time-consuming, nevertheless. Numerous applications utilizing deep learning to identify diseased plants and improve detection rates have circumvented the issue of model parameter size, which was previously a significant obstacle. This study aims to introduce a hybrid deep learning model that is specifically developed for the automated detection of diseases on tomato foliage. The recommended framework comprises Segmentation, Pre-processing, Extraction, and synthesis of features establishment of classification. To create more potent features, use the Fusion technique in this case by merging deep features from two well-known models (DenseNet121 and EfficientNet B7). This study employed the data augmentation technique to increase the quantity and variety of images utilized in the model's training process to enable it to comprehend more intricate scenarios within the dataset. The proposed model shows an accuracy of 98.55%, which detects the lesions with a precision of 97.48%, recall 97.8%, and F1-score 96.47%. This degree of accuracy demonstrates how effectively our approach detects the presence of illnesses on tomato plant leaves.</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

TLD-FDL: Tomato Leaf Disease Classification Using End-to-End Fusion of Deep Learning Models

  • 期刊论文

  • 发表日期：

    2025年10月30日

查看全文
  • Vishal kumar,
  • Manpreet Kaur,
  • Vikas Mittal,
  • Taghreed Alqaisi,
  • Vivek Bhardwaj

摘要

Plant pathogens pose a significant risk to the security of food supplies worldwide. Plant disease identification is still difficult and time-consuming, nevertheless. Numerous applications utilizing deep learning to identify diseased plants and improve detection rates have circumvented the issue of model parameter size, which was previously a significant obstacle. This study aims to introduce a hybrid deep learning model that is specifically developed for the automated detection of diseases on tomato foliage. The recommended framework comprises Segmentation, Pre-processing, Extraction, and synthesis of features establishment of classification. To create more potent features, use the Fusion technique in this case by merging deep features from two well-known models (DenseNet121 and EfficientNet B7). This study employed the data augmentation technique to increase the quantity and variety of images utilized in the model's training process to enable it to comprehend more intricate scenarios within the dataset. The proposed model shows an accuracy of 98.55%, which detects the lesions with a precision of 97.48%, recall 97.8%, and F1-score 96.47%. This degree of accuracy demonstrates how effectively our approach detects the presence of illnesses on tomato plant leaves.