Scholarlink

<p>In modern farming, Artificial Intelligence has the power to change things by making it easier to quickly and accurately identify crop illnesses. Tomatoes are a very important part of agriculture around the world, but they are quite prone to diseases that can have a big impact on quality and output. This paper presents a deep learning strategy for automatically classifying tomato illnesses and deficits using a finely tuned ResNet-50 model. Our architecture uses advanced training methods like label smoothing, mixed-precision training, and adaptive learning rate scheduling, together with effective data augmentation methods to improve generalization. The model was trained using the Tomato Village dataset, which has eight classes of diseases and deficits. It got a test accuracy of 95.66% and showed strong class-wise precision and recall. This method makes it possible to monitor the health of crops in real time with a system that can be scaled up, is reliable, and can be used in the field. This helps reduce crop losses and promote sustainable farming.</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

Neural Network-Based Diagnostic Systems for Tomato Leaf Diseases in Smart Farming Environments

  • 期刊论文

  • 发表日期：

    2026年4月17日

查看全文
  • Sheenam,
  • Kanika Sharma

摘要

In modern farming, Artificial Intelligence has the power to change things by making it easier to quickly and accurately identify crop illnesses. Tomatoes are a very important part of agriculture around the world, but they are quite prone to diseases that can have a big impact on quality and output. This paper presents a deep learning strategy for automatically classifying tomato illnesses and deficits using a finely tuned ResNet-50 model. Our architecture uses advanced training methods like label smoothing, mixed-precision training, and adaptive learning rate scheduling, together with effective data augmentation methods to improve generalization. The model was trained using the Tomato Village dataset, which has eight classes of diseases and deficits. It got a test accuracy of 95.66% and showed strong class-wise precision and recall. This method makes it possible to monitor the health of crops in real time with a system that can be scaled up, is reliable, and can be used in the field. This helps reduce crop losses and promote sustainable farming.