Scholarlink

This research paper explores the advancement of artificial intelligence (AI) to recognize diseases from pictures of disease-ridden plants. The framework leverages progressed or advanced image processing and deep learning advanced learning techniques within a federated learning environment, and this system aims to provide farmers with an accessible, real-time diagnostic tool directly on their mobile devices. Experimental results shows that the proposed federated learning custom model obtains 92.45% accuracy, outperforming MobileNet, VGG16 (78.27%) and VGG19 (74.09%). While Vision Transformer Models having higher accuracy, but do the extremely high computational resources, making them unsuitable for real-world agricultural applications. The intelligent model ought to identify the agricultural diseases early by recognizing the side effects of infections in crop pictures, in this manner expanding yields and diminishing the collapse of labor cultivate work. This approach rises above traditional, labour-intensive monitoring approaches and provides large-scale, cost-effective (efficient) solutions to important environmental problems. It can greatly improve plant disease detection keeping data privacy intact. Future work will focus on optimizing the computational costings and integration of more crop species.

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

Advancements in Agriculture Disease Detection: Employing Federated Learning for Plant Leaves

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Umesh Gupta,
  • Rajiv Kumar,
  • Akshat Jain,
  • Ayushman Pranav,
  • Sakshi Gupta

摘要

This research paper explores the advancement of artificial intelligence (AI) to recognize diseases from pictures of disease-ridden plants. The framework leverages progressed or advanced image processing and deep learning advanced learning techniques within a federated learning environment, and this system aims to provide farmers with an accessible, real-time diagnostic tool directly on their mobile devices. Experimental results shows that the proposed federated learning custom model obtains 92.45% accuracy, outperforming MobileNet, VGG16 (78.27%) and VGG19 (74.09%). While Vision Transformer Models having higher accuracy, but do the extremely high computational resources, making them unsuitable for real-world agricultural applications. The intelligent model ought to identify the agricultural diseases early by recognizing the side effects of infections in crop pictures, in this manner expanding yields and diminishing the collapse of labor cultivate work. This approach rises above traditional, labour-intensive monitoring approaches and provides large-scale, cost-effective (efficient) solutions to important environmental problems. It can greatly improve plant disease detection keeping data privacy intact. Future work will focus on optimizing the computational costings and integration of more crop species.