Adaptive Agriculture System with Soil Health Prediction using Ensemble Techniques and Smart Irrigation using Deep Learning

Agriculture is a cornerstone of the country’s economy, serving as a fundamental necessity for human survival and a critical driver of socio-economic stability. In traditional farming, the efficiency of crop yield is often limited, and resources are not managed effectively, leading to a significant reduction in overall productivity. This paper introduces an IoT-based adaptive agriculture system using a real-time soil health prediction algorithm, where it leverages the Custom Stacked Ensemble model combined with K-Fold Stratified cross-validation to generate predictions from dynamic environmental and soil conditions. The proposed model achieves an accuracy of 94.5%. Additionally, the system incorporates a smart irrigation mechanism that regulates water delivery to crops, ensuring optimal growth while conserving water resources. The irrigation process is automated using the Enhanced GRU model, which achieves an accuracy of 99.43%, further enhancing efficiency and sustainability. The system collects the data from the sensor in the field via a LoRa-based network that communicates with each other to transmit the data to the IoT server. By combining the weather forecasts and the data obtained from the sensors, the system gives soil suitability for the farmer and also predicts the precise irrigation levels and soil fertility, aiding adaptive farming.