Design and implementation of an IoT-based system for intelligent crop health monitoring

Abstract

This paper presents the development of an intelligent IoT device for automated, real-time monitoring of crop conditions in agriculture. The proposed solution involves Raspberry Pi Zero 2 W hardware, multi-sensor modules for environmental data collection, NB-IoT for long-range wireless communication, and the YOLOv8 convolutional neural network for plant image analysis. The objective is to create a compact, low-cost, and energy-efficient solution that enables early detection of plant diseases and environmental stress in remote or infrastructure-poor agricultural areas. The developed system enables accurate identification of disease symptoms and damage on crop leaves based on visual and environmental input, facilitating timely intervention and reducing yield loss. The YOLOv8 model was adapted for resource-constrained edge deployment, trained on a custom dataset of strawberry leaf diseases, and integrated into the embedded device with high accuracy and low latency. System testing confirmed reliable performance under field conditions, with successful image classification and robust NB-IoT communication. The proposed solution is scalable and applicable to various crops and contributes to the practical implementation of precision agriculture and intelligent farming systems.