Scholarlink

<p>In modern agriculture, the detection of plant diseases is crucial for enhancing crop productivity. Predicting disease onset and providing advice to farmers are essential steps to achieve increased yields on a large scale. The research addresses the critical need for timely and accurate plant leaf disease diagnosis to prevent growth issues. Leveraging deep learning advancements, the work confronts challenges like small lesion characteristics, distorted backgrounds, data imbalances, and limited generalization in agricultural datasets. After preprocessing the leaf images with tasks like data augmentation and resizing, a sheaf attention U-net with K-means clustering (SAUKC) is employed for segmentation to identify the region of interest. The segmented features are then input into the Orientation-guided Crystal Edge Deep Network (OCEDN) for infection detection. Fine-tuning with the improved kookaburra optimization algorithm (IKOA) addresses training challenges. The proposed method accurately identifies plant leaf diseases, achieving a remarkable accuracy rate of 98%. The validity of the statistical analysis is confirmed to substantiate the outcomes regarding accuracy, specificity, and recall.</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

Innovative Deep Learning Framework for Accurate Plant Disease Detection and Crop Productivity Enhancement

  • 期刊论文

  • 发表日期：

    2025年1月30日

查看全文
  • Mohan M.,
  • S. Anandamurugan

摘要

In modern agriculture, the detection of plant diseases is crucial for enhancing crop productivity. Predicting disease onset and providing advice to farmers are essential steps to achieve increased yields on a large scale. The research addresses the critical need for timely and accurate plant leaf disease diagnosis to prevent growth issues. Leveraging deep learning advancements, the work confronts challenges like small lesion characteristics, distorted backgrounds, data imbalances, and limited generalization in agricultural datasets. After preprocessing the leaf images with tasks like data augmentation and resizing, a sheaf attention U-net with K-means clustering (SAUKC) is employed for segmentation to identify the region of interest. The segmented features are then input into the Orientation-guided Crystal Edge Deep Network (OCEDN) for infection detection. Fine-tuning with the improved kookaburra optimization algorithm (IKOA) addresses training challenges. The proposed method accurately identifies plant leaf diseases, achieving a remarkable accuracy rate of 98%. The validity of the statistical analysis is confirmed to substantiate the outcomes regarding accuracy, specificity, and recall.