Scholarlink

The agriculture industry greatly aids a nation’s economic maturation. Although, plant diseases are the main factor causing crop devastation, which lowers the amount and quality of agricultural production and results in significant financial loss. Therefore, it is essential to detect plant illnesses well in advance to prevent crop devastation. Automated and timely identification of these illnesses will boost output and facilitate system digitization. Our primary goal is to identify potato diseases using sophisticated machine-learning algorithms on leaf image data. This paper uses image-based Convolutional Neural Network (CNN) models with a transfer learning strategy to study and compare performance using pre-trained models via VGG16, ResNet50, and MobileNetV2. In this work, our primary focus is on potato diseases consists of 2152 images of three classes: Early Blight, Late Blight and Healthy. It is ascertained through experimentation that MobileNetV2 outperformed other models. The proposed method achieved a 90.71%, 65.89% and 93.27% accuracy for potato leaf images. This model helps the farmer detect disease early and help them in taking preventative action well on time to reduce their loss.

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

Transfer Learning for Leaf Disease Image Classification via CNN Model

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Kirti Dabas,
  • Anuradha Chug,
  • Amit Prakash Singh,
  • Ravinder Kumar,
  • Veerubommu Shanumugam

摘要

The agriculture industry greatly aids a nation’s economic maturation. Although, plant diseases are the main factor causing crop devastation, which lowers the amount and quality of agricultural production and results in significant financial loss. Therefore, it is essential to detect plant illnesses well in advance to prevent crop devastation. Automated and timely identification of these illnesses will boost output and facilitate system digitization. Our primary goal is to identify potato diseases using sophisticated machine-learning algorithms on leaf image data. This paper uses image-based Convolutional Neural Network (CNN) models with a transfer learning strategy to study and compare performance using pre-trained models via VGG16, ResNet50, and MobileNetV2. In this work, our primary focus is on potato diseases consists of 2152 images of three classes: Early Blight, Late Blight and Healthy. It is ascertained through experimentation that MobileNetV2 outperformed other models. The proposed method achieved a 90.71%, 65.89% and 93.27% accuracy for potato leaf images. This model helps the farmer detect disease early and help them in taking preventative action well on time to reduce their loss.