An Embedded AI System for Automated Crop irrigation and pest Monitoring

Abstract

Modern agriculture is rapidly adopting Artificial Intelligence (AI) and Internet of Things (IoT) technologies to improve crop monitoring and decision-making. Many existing systems focus either on water stress detection or pest detection separately. The proposed system integrates both functions into a single platform. It uses a camera module and environmental sensors connected to a Raspberry Pi (5/4) as the main controller. A Convolutional Neural Network (CNN) model processes leaf images captured by the AI camera, while a soil moisture sensor supports water stress analysis. The system classifies crops into three categories: healthy, water-stressed, and pest-infected. Based on the output, it provides real-time recommendations for irrigation and pesticide application. This reduces manual inspection, prevents unnecessary chemical usage, saves water, and improves crop productivity.

Code and data availability

The paper's CNN phenotyping/classification analysis is built directly on two public Kaggle image datasets (PlantVillage plant disease and Crop Water Stress), explicitly cited with URLs. No author code or trained model is deposited.