PlantOS: An IoT-Enabled Smart Greenhouse Monitoring System with Integrated Machine Learning for Green Gram Cultivation
Abstract
Precision agriculture calls for cost-effective intelligent devices to monitor crops efficiently and perform irrigation. In this study, we introduce PlantOS, a smart greenhouse mung bean (Vigna radiata) crop monitoring system enabled by the Internet of Things (IoT). The proposed system combines a spectrometer AS7341 sensor with soil moisture, temperature, and pressure sensors using an ESP32 sensor node interfaced with the FastAPI backend server. Three models based on the random forest algorithm are designed to predict seed germination, classify plant nutrient type Nitrogen, phosphorus, and potassium (NPK), and control irrigation. A Streamlit application is highly practical since it allows us to monitor the process in real-time. Moreover, our research features a Groq LLaMA-powered assistant providing us with relevant recommendations about plant maintenance. Such an assistant provides valuable suggestions related to the particular case. As shown in our experiments, this system can accurately forecast germination with an F1 score of 91.6 percent. Moreover, it enables efficient management of plant watering with high reliability. Overall, this work demonstrates the effectiveness of integrating light and machine learning technologies into plant maintenance strategies within greenhouses at low costs.