Scholarlink

<p>The increasing adoption of IoT-based Smart Farming (SF) has transformed agriculture, enhancing productivity and efficiency. However, challenges such as cybersecurity risks, rural connectivity issues, and device interoperability hinder its full potential. This review systematically examines 88 studies published since 2019, focusing on key aspects such as application domains, privacy and security, communication protocols, sensors, and devices. Evaluations using datasets like CropDeep, regional agricultural data, and precision agriculture datasets demonstrate the effectiveness of deep learning (DL) and machine learning (ML) approaches in crop recommendation, disease detection, and yield prediction. Efficient MobileNet (EffiMob-Net) achieves a 99.92% accuracy rate in disease identification, while hybrid optimization algorithms improve IoT node placement by 18%, enhancing energy efficiency and coverage. The comparative analysis highlights DL models outperforming ML, with CNN achieving 99.81% accuracy in plant disease detection. Additionally, AI-based solutions, including forest cover change detection and DL-driven yield prediction, show improvements of up to 22% in forecasting accuracy. Scientific mapping identifies gaps in IoT security, dataset quality, and standardization. To address existing research limitations, a 5G-based SF framework is proposed, aiming to improve connectivity, real-time data processing, and automation in SF. This review advocates for robust security measures, high-quality datasets, and AI-driven innovations to drive future advancements in precision agriculture.</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

Smart Farming Revolution: A Cutting-Edge Review of Deep Learning and IoT Innovations in Agriculture

  • 期刊论文

  • 发表日期：

    2025年3月14日

查看全文
  • J. Siva Prashanth,
  • G. Bala Krishna,
  • A. V. Krishna Prasad,
  • P. Ravinder Rao

摘要

The increasing adoption of IoT-based Smart Farming (SF) has transformed agriculture, enhancing productivity and efficiency. However, challenges such as cybersecurity risks, rural connectivity issues, and device interoperability hinder its full potential. This review systematically examines 88 studies published since 2019, focusing on key aspects such as application domains, privacy and security, communication protocols, sensors, and devices. Evaluations using datasets like CropDeep, regional agricultural data, and precision agriculture datasets demonstrate the effectiveness of deep learning (DL) and machine learning (ML) approaches in crop recommendation, disease detection, and yield prediction. Efficient MobileNet (EffiMob-Net) achieves a 99.92% accuracy rate in disease identification, while hybrid optimization algorithms improve IoT node placement by 18%, enhancing energy efficiency and coverage. The comparative analysis highlights DL models outperforming ML, with CNN achieving 99.81% accuracy in plant disease detection. Additionally, AI-based solutions, including forest cover change detection and DL-driven yield prediction, show improvements of up to 22% in forecasting accuracy. Scientific mapping identifies gaps in IoT security, dataset quality, and standardization. To address existing research limitations, a 5G-based SF framework is proposed, aiming to improve connectivity, real-time data processing, and automation in SF. This review advocates for robust security measures, high-quality datasets, and AI-driven innovations to drive future advancements in precision agriculture.