Scholarlink

The proposed research aims at integrating UAV images with deep learning models for monitoring and analyzing the health of a wheat crop concerning common diseases of wheat, which include stripe rust, stem rust, head scab, and fusarium head blight. With the adoption of precision agriculture systems, there is a need to detect diseases as accurately and on time as possible to ensure maximum yields and overall crop health. Traditional methods to detect diseases in plants are more labor-intensive, leading to some delays in implementing the intervention required. The new system of automating disease identification would be worthwhile because it offers higher accuracy with quicker results that allow for earlier decision-making and focused treatment. In this research study, several architectures of deep learning have been investigated: DenseNet-50, ResNet-121, Inception V3, and EfficientNet V4 for their capability of wheat disease image classification from the drone-captured images. It indicates that DenseNet-50 achieves the highest accuracy and lowest misclassification rates when compared to the other models; the second one is ResNet-121, followed by Inception V3 and EfficientNet V4. These results imply that DenseNet-50 may be deployed for real-time deployment in agricultural scenarios with significant improvement in disease detection, pesticide recommendation, and nutrient management. The research promises to combine UAV technology with deep learning to enhance crop health management, reduce labor costs, and optimize resource use in agriculture, thus contributing to sustainable farming practices and increased productivity.

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

Integrating UAV Imagery and Deep Learning Models for Automated Crop Health Monitoring and Analysis

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Anugrah Srivastava,
  • Jagendra Singh,
  • Monika Dandotiya,
  • Pongkit Ekvitayavetchanukul,
  • Manoj Rana,
  • Bakshish Singh

摘要

The proposed research aims at integrating UAV images with deep learning models for monitoring and analyzing the health of a wheat crop concerning common diseases of wheat, which include stripe rust, stem rust, head scab, and fusarium head blight. With the adoption of precision agriculture systems, there is a need to detect diseases as accurately and on time as possible to ensure maximum yields and overall crop health. Traditional methods to detect diseases in plants are more labor-intensive, leading to some delays in implementing the intervention required. The new system of automating disease identification would be worthwhile because it offers higher accuracy with quicker results that allow for earlier decision-making and focused treatment. In this research study, several architectures of deep learning have been investigated: DenseNet-50, ResNet-121, Inception V3, and EfficientNet V4 for their capability of wheat disease image classification from the drone-captured images. It indicates that DenseNet-50 achieves the highest accuracy and lowest misclassification rates when compared to the other models; the second one is ResNet-121, followed by Inception V3 and EfficientNet V4. These results imply that DenseNet-50 may be deployed for real-time deployment in agricultural scenarios with significant improvement in disease detection, pesticide recommendation, and nutrient management. The research promises to combine UAV technology with deep learning to enhance crop health management, reduce labor costs, and optimize resource use in agriculture, thus contributing to sustainable farming practices and increased productivity.