Scholarlink

<p>Chili is a vital crop in India, widely used in food and medicine, yet its productivity is increasingly threatened by diverse leaf diseases that reduce yield and quality. Traditional identification methods rely heavily on visual inspection, which is timeconsuming and often inaccurate, leading to delayed intervention and significant crop losses. To address this research gap, a novel framework, Conditional Orangutan Optimization Algorithm based FractalCovNet (COOA_FractalCovNet), is proposed for early and accurate chili leaf disease detection. The approach integrates Kuwahara filtering for noise reduction, watershedbased segmentation for precise leaf extraction, and data augmentation to enhance variability. Disease detection is performed using FractalCovNet optimized with COOA, enabling robust feature learning. Experimental results demonstrate that COOA_FractalCovNet outperforms existing methods, achieving an accuracy of 91.772%, a true positive rate of 90.651%, and a true negative rate of 92.917%. These outcomes confirm the model’s effectiveness in providing a reliable, scalable, and timely solution for sustainable chili crop management.</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

Conditional Orangutan Optimization Algorithm Based FractalCovNet for Chili Leaf Disease Detection

  • 期刊论文

  • 发表日期：

    2025年8月13日

查看全文
  • R. Thyagaraj,
  • George Glan Devadhas,
  • T. Y. Satheesha

摘要

Chili is a vital crop in India, widely used in food and medicine, yet its productivity is increasingly threatened by diverse leaf diseases that reduce yield and quality. Traditional identification methods rely heavily on visual inspection, which is timeconsuming and often inaccurate, leading to delayed intervention and significant crop losses. To address this research gap, a novel framework, Conditional Orangutan Optimization Algorithm based FractalCovNet (COOA_FractalCovNet), is proposed for early and accurate chili leaf disease detection. The approach integrates Kuwahara filtering for noise reduction, watershedbased segmentation for precise leaf extraction, and data augmentation to enhance variability. Disease detection is performed using FractalCovNet optimized with COOA, enabling robust feature learning. Experimental results demonstrate that COOA_FractalCovNet outperforms existing methods, achieving an accuracy of 91.772%, a true positive rate of 90.651%, and a true negative rate of 92.917%. These outcomes confirm the model’s effectiveness in providing a reliable, scalable, and timely solution for sustainable chili crop management.