Intelligent Crop Yield Prediction Using IoT and Optimized Deep Learning for Precision Agriculture

The accurate prediction of crop yield plays an important role in sustainable agriculture. Traditional prediction methods depend on historical data and general environmental assumptions. These methods lack accuracy and fail to adapt to real-time field conditions. The proposed research develops an intelligent system that integrates IoT sensors with an optimized deep-learning model for precise yield prediction. The system collects real-time data using IoT sensors that measure soil moisture, temperature, humidity, pH, and light intensity. A cloud-based platform processes this data with a MobileNet-Inspired LSTM model. A self-adaptive Adam optimizer enhances training efficiency by adjusting learning rates dynamically. The model analyzes real-time environmental conditions along with future weather data from an external API. A web dashboard presents predictions to users, allowing better decision-making. The proposed method solves key challenges in yield prediction by improving data accuracy. The integration of IoT ensures real-time monitoring. The MobileNet-Inspired LSTM model improves prediction accuracy while maintaining efficiency. The self-adaptive Adam optimizer accelerates training, reducing errors in yield estimation. The system is tested on coconut, rice, guava, and papaya yield datasets. The results show that the proposed model achieves an R-squared score above 0.90 for most crops, indicating strong predictive performance. Compared to existing methods, the proposed approach reduces mean absolute percentage error (MAPE) and root mean square error (RMSE), proving its reliability. The research enhances precision agriculture by providing an accurate, real-time, and scalable crop yield prediction system.