Application of IoT and Machine Learning to Control and Monitor the Wellbeing of Plants in a Green House

Abstract

The increasing impact of climate variability has necessitated the development of intelligent systems for efficient greenhouse management. This study presents the design and implementation of a smart greenhouse that integrates Internet of Things (IoT) sensors with machine learning models for real-time monitoring, prediction, and control of crop health. The system comprises a greenhouse structure with frame and canopy, IoT-based environmental sensing units, a data logger, and an automated irrigation system including a water pump, overhead tank, and pipe network. Environmental data—temperature, humidity, soil moisture, CO₂ concentration, and crop growth level—were collected at 3-hour intervals between January 15 and March 13, 2026, yielding over 400 observations. Crop health was visually classified into Good, Moderate, and Bad categories and used as the target variable for supervised learning. A comparative analysis of Decision Tree, Support Vector Machine (SVM), Artificial Neural Network (ANN), and Random Forest models was conducted. The Random Forest model achieved the highest prediction performance with an accuracy of 84.29%, precision of 84.06%, recall of 84.29%, and F1-score of 84.15%, outperforming ANN (82.10%), SVM (80.47%), and Decision Tree (78.65%). The integration of the predictive model with a rule-based irrigation system resulted in a 28% reduction in water usage compared to conventional manual irrigation, while maintaining over 84% accuracy in crop health prediction. The system enabled timely intervention in over 90% of potential plant stress conditions. Real-time monitoring and control were achieved through a mobile application dashboard. The results demonstrate significant improvements in water use efficiency, predictive accuracy, and overall greenhouse management, resource utilization, and enhancing sustainability in precision agriculture.