Scholarlink

Potato stands as a vital staple food worldwide, with its consumption steadily rising and becoming the fourth most consumed staple food globally. However, the prevalence of potato diseases has posed a significant challenge, impairing both the quality and quantity of harvests. Timely and accurate disease classification is imperative for effective disease management and crop protection. Leveraging recent advancements in deep learning and computer vision, this study provides a thorough analysis and evaluation of the differences between of prominent convolutional neural network architectures, namely VGG16, VGG19, ResNet50, ResNet152, and InceptionV3, for classifying three distinct disease classes in potato plants: early blight, late blight, and healthy. In the investigation of a dataset comprising 2152 images, this study scrutinizes the performance metrics of diverse models. The results underscore the remarkable efficiency of InceptionV3, demonstrating an impressive 91.84% accuracy in classifying potato diseases. Significantly, the incorporation of deep learning approaches presents a potential solution for tackling the complex challenges linked to identifying and categorizing potato diseases. Additionally, the paper underscores the vital significance of automated disease detection in preserving worldwide potato production, ensuring food security, and promoting sustainable agricultural practices.

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

Automated Potato Disease Classification Using Deep Learning - A Comparative Analysis of Convolutional Neural Networks

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Swati Pandey,
  • Mayuri Gupta,
  • Ashish Mishra,
  • Ashutosh Mishra,
  • Jayesh Gangrade

摘要

Potato stands as a vital staple food worldwide, with its consumption steadily rising and becoming the fourth most consumed staple food globally. However, the prevalence of potato diseases has posed a significant challenge, impairing both the quality and quantity of harvests. Timely and accurate disease classification is imperative for effective disease management and crop protection. Leveraging recent advancements in deep learning and computer vision, this study provides a thorough analysis and evaluation of the differences between of prominent convolutional neural network architectures, namely VGG16, VGG19, ResNet50, ResNet152, and InceptionV3, for classifying three distinct disease classes in potato plants: early blight, late blight, and healthy. In the investigation of a dataset comprising 2152 images, this study scrutinizes the performance metrics of diverse models. The results underscore the remarkable efficiency of InceptionV3, demonstrating an impressive 91.84% accuracy in classifying potato diseases. Significantly, the incorporation of deep learning approaches presents a potential solution for tackling the complex challenges linked to identifying and categorizing potato diseases. Additionally, the paper underscores the vital significance of automated disease detection in preserving worldwide potato production, ensuring food security, and promoting sustainable agricultural practices.