Scholarlink

<p>Effective protection against agricultural diseases and pests, which cause significant production losses, is essential for ensuring global food security. Conventional methods often employ separate deep learning models for plant disease detection and pest classification, leading to increased computational overhead, scalability issues, and inefficiencies that hinder real-world deployment. Additionally, traditional manual inspections, despite their widespread use, remain time-consuming, subjective, and prone to human errors, limiting their practicality for large-scale agricultural applications. To address these challenges, this study proposes a hybrid deep learning model that fuses DenseNet-121 and ResNet-50 to simultaneously perform crop disease detection and pest classification, enhancing efficiency while reducing computational complexity. The proposed system achieves state-of-the-art accuracy of 99.13% in disease detection and 98.67% in pest classification after being trained on the Subset of IP 102 dataset and the New Plant Village Dataset. Scalability is improved through advanced preprocessing techniques, such as data augmentation and image enhancement, facilitating efficient real-time deployment. Despite its high accuracy, the system’s performance may be influenced by lighting variations and image resolution, while its computational demands could pose challenges in low-resource environments. Future research will focus on integrating IoT-enabled drones for autonomous monitoring, optimizing edge computing, and expanding datasets. This fusion-based approach provides a scalable and sustainable solution for precision farming and global food security, representing a significant advancement in AI-driven agricultural diagnostics.</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

Towards automated crop protection: fusion of densenet121-resnet50 model for disease detection and pest recognition

  • 期刊论文

  • 发表日期：

    2025年9月17日

查看全文
  • Vedansh Sood,
  • Shiv Shankar Prasad Shukla,
  • Anil Kumar Yadav,
  • Sparsh Tiwari,
  • Yashashvi Srivastava

摘要

Effective protection against agricultural diseases and pests, which cause significant production losses, is essential for ensuring global food security. Conventional methods often employ separate deep learning models for plant disease detection and pest classification, leading to increased computational overhead, scalability issues, and inefficiencies that hinder real-world deployment. Additionally, traditional manual inspections, despite their widespread use, remain time-consuming, subjective, and prone to human errors, limiting their practicality for large-scale agricultural applications. To address these challenges, this study proposes a hybrid deep learning model that fuses DenseNet-121 and ResNet-50 to simultaneously perform crop disease detection and pest classification, enhancing efficiency while reducing computational complexity. The proposed system achieves state-of-the-art accuracy of 99.13% in disease detection and 98.67% in pest classification after being trained on the Subset of IP 102 dataset and the New Plant Village Dataset. Scalability is improved through advanced preprocessing techniques, such as data augmentation and image enhancement, facilitating efficient real-time deployment. Despite its high accuracy, the system’s performance may be influenced by lighting variations and image resolution, while its computational demands could pose challenges in low-resource environments. Future research will focus on integrating IoT-enabled drones for autonomous monitoring, optimizing edge computing, and expanding datasets. This fusion-based approach provides a scalable and sustainable solution for precision farming and global food security, representing a significant advancement in AI-driven agricultural diagnostics.