Scholarlink

<p>Detecting plant diseases is essential for maintaining crop health and maximizing agricultural productivity. While recent advancements in machine learning and deep learning methods for plant disease diagnosis show significant promise, they also present challenges—particularly related to data annotation and the limitations of task-specific models, which often struggle to generalize across different plant diseases. Traditional methods face additional obstacles due to their dependence on crop-specific models and the need for manual inspections, resulting in inefficiencies and limited scalability. This study proposes a generalized approach for efficiently identifying a wide range of plant diseases. By integrating patch-based analysis with the self-supervised learning technique SimCLR, the method enables farmers and researchers to detect unhealthy leaves across various crop species efficiently. To evaluate the effectiveness of our approach, we trained and tested the model using the well-regarded PlantVillage dataset, known for its extensive and diverse representation. Our approach achieved an accuracy of <b>97.57%</b>. Moreover, when tested on novel datasets, the model achieved an accuracy of <b>99.22%</b>, demonstrating its robustness and strong generalization capability to previously unseen data.</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

A generalized novel approach for plant disease detection based on SimCLR and patch-based analysis

  • 期刊论文

  • 发表日期：

    2025年8月14日

查看全文
  • Mohamed T. Faheem Saidahmd,
  • Reda M. Elbasiony,
  • Marwa R. M. Bastwesy,
  • Asmaa A. M. Hagar

摘要

Detecting plant diseases is essential for maintaining crop health and maximizing agricultural productivity. While recent advancements in machine learning and deep learning methods for plant disease diagnosis show significant promise, they also present challenges—particularly related to data annotation and the limitations of task-specific models, which often struggle to generalize across different plant diseases. Traditional methods face additional obstacles due to their dependence on crop-specific models and the need for manual inspections, resulting in inefficiencies and limited scalability. This study proposes a generalized approach for efficiently identifying a wide range of plant diseases. By integrating patch-based analysis with the self-supervised learning technique SimCLR, the method enables farmers and researchers to detect unhealthy leaves across various crop species efficiently. To evaluate the effectiveness of our approach, we trained and tested the model using the well-regarded PlantVillage dataset, known for its extensive and diverse representation. Our approach achieved an accuracy of 97.57%. Moreover, when tested on novel datasets, the model achieved an accuracy of 99.22%, demonstrating its robustness and strong generalization capability to previously unseen data.