Scholarlink

<p>Plant diseases pose a significant threat to agricultural productivity, often leading to severe crop losses. Early, precise detection and classification using computerized image processing can help mitigate damage, reducing losses and boosting yields. This paper presents Hy-SALDD (Hybrid Segmentation-based Agricultural Leaf Disease Detection), an innovative approach for identifying plant leaf diseases. The method integrates enhanced segmentation, feature extraction, and classification techniques. A hybrid segmentation model combining improved PSP-Net and enhanced HED efficiently isolates diseased regions in leaf images. Feature extraction incorporates deep features from the segmentation model and texture features using LBP. Feature selection is optimized through the Black Widow Optimization (BWO) technique, ensuring a compact and discriminative feature set. Classification is performed using a Bayesian-optimized SVM, improving detection accuracy. The proposed framework demonstrates significant potential for precise and efficient plant disease diagnosis.</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

Hybrid segmentation-based agricultural leaf disease detection (Hy-SALDD) using black widow optimization for feature selection, and Bayesian-optimized SVM classification

  • 期刊论文

  • 发表日期：

    2025年3月12日

查看全文
  • Aaditya Jain,
  • Rakesh Kumar Dwivedi

摘要

Plant diseases pose a significant threat to agricultural productivity, often leading to severe crop losses. Early, precise detection and classification using computerized image processing can help mitigate damage, reducing losses and boosting yields. This paper presents Hy-SALDD (Hybrid Segmentation-based Agricultural Leaf Disease Detection), an innovative approach for identifying plant leaf diseases. The method integrates enhanced segmentation, feature extraction, and classification techniques. A hybrid segmentation model combining improved PSP-Net and enhanced HED efficiently isolates diseased regions in leaf images. Feature extraction incorporates deep features from the segmentation model and texture features using LBP. Feature selection is optimized through the Black Widow Optimization (BWO) technique, ensuring a compact and discriminative feature set. Classification is performed using a Bayesian-optimized SVM, improving detection accuracy. The proposed framework demonstrates significant potential for precise and efficient plant disease diagnosis.