AI and IoT-powered edge device optimized for crop pest and disease detection

Area of Science:

• Agricultural Technology
• Computer Science
• Environmental Science

Background:

• Climate change intensifies pest and disease outbreaks in cereal crops, causing significant yield losses.
• Existing monitoring technologies often lack portability, cost-effectiveness, and energy efficiency for resource-constrained agricultural settings.
• There is a critical need for edge-compatible solutions for real-time crop health monitoring.

Purpose of the Study:

• To develop a portable, cost-effective, and energy-efficient smart IoT device for early pest and disease detection in crops.
• To integrate a lightweight, explainable convolutional neural network (CNN) optimized for edge applications.
• To provide a practical tool for farmers to enhance crop health management and food security.

Main Methods:

• Development of a portable smart IoT device featuring a high-definition camera and Raspberry Pi 5.
• Integration of a novel lightweight CNN, Tiny-LiteNet, for edge-based image processing and analysis.
• Utilization of a GSM/GPRS module for seamless cloud communication and data transmission.

Main Results:

• Tiny-LiteNet achieved high accuracy (98.6%), F1-score (98.4%), and Recall (98.2%) with a fast inference time (80 ms).
• The model boasts a compact size (1.2 MB, 1.48 million parameters), outperforming traditional CNNs in efficiency for edge computing.
• The smart device demonstrated low power consumption and a user-friendly design suitable for agricultural field use.

Conclusions:

• The developed smart IoT device offers a practical and efficient solution for real-time pest and disease detection in agriculture.
• Tiny-LiteNet's performance and efficiency make it ideal for edge AI applications in resource-constrained environments.
• This technology supports sustainable agriculture practices, enhances food security, and empowers farmers with actionable insights.