Scholarlink

In the quest for more food production to feed the booming population of the modern world, maintaining plant health is critical to ensuring global food security. In this regard, one important field of study is the early and precise identification of plant diseases. Artificial intelligence (AI) and deep learning approaches, in particular, have demonstrated encouraging advances in this subject in recent years. Using the “A Database of Leaf Images: Practice towards Plant Conservation with Plant Pathology” dataset, this study explores the use of deep learning-based methods for the diagnosis of plant diseases. The research evaluates the effectiveness of well-known deep transfer learning models, including VGG16, GoogleNet, ResNet50, and DarkNet53, in correctly sorting leaf images into healthy and unhealthy categories. The results showed great promise, especially for DarkNet53, which achieved an accuracy of 99.7%. VGG16 and ResNet50 followed with 97% and 90% accuracy, respectively. Through the provision of a unique approach to early disease diagnosis, assistance in maintaining crop health and reduction of agricultural waste, these findings contribute to sustainability. By using cutting-edge deep learning technology to potentially improve food security, promote human health, foster agricultural technological advancement, encourage sustainable production practices, and support climate adaptation efforts, the current study is said to be in line with Sustainable Development Goals (SDGs) such as Zero Hunger, Good Health and Well-Being, Industry, Innovation, and Infrastructure, Responsible Consumption and Production, and Climate Action.

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

Evaluation of State-of-the-Art Models for Advancing Plant Disease Diagnosis Through Deep Learning: A Sustainable Approach

  • 会议论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年

查看全文
  • Masud Kabir,
  • Sami Ekici

摘要

In the quest for more food production to feed the booming population of the modern world, maintaining plant health is critical to ensuring global food security. In this regard, one important field of study is the early and precise identification of plant diseases. Artificial intelligence (AI) and deep learning approaches, in particular, have demonstrated encouraging advances in this subject in recent years. Using the “A Database of Leaf Images: Practice towards Plant Conservation with Plant Pathology” dataset, this study explores the use of deep learning-based methods for the diagnosis of plant diseases. The research evaluates the effectiveness of well-known deep transfer learning models, including VGG16, GoogleNet, ResNet50, and DarkNet53, in correctly sorting leaf images into healthy and unhealthy categories. The results showed great promise, especially for DarkNet53, which achieved an accuracy of 99.7%. VGG16 and ResNet50 followed with 97% and 90% accuracy, respectively. Through the provision of a unique approach to early disease diagnosis, assistance in maintaining crop health and reduction of agricultural waste, these findings contribute to sustainability. By using cutting-edge deep learning technology to potentially improve food security, promote human health, foster agricultural technological advancement, encourage sustainable production practices, and support climate adaptation efforts, the current study is said to be in line with Sustainable Development Goals (SDGs) such as Zero Hunger, Good Health and Well-Being, Industry, Innovation, and Infrastructure, Responsible Consumption and Production, and Climate Action.