ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING APPLICATIONS IN PRECISION AGRICULTURE: ENHANCING CROP YIELD PREDICTION, DISEASE DETECTION, AND RESOURCE OPTIMIZATION

Abstract

Precision agriculture has become more dependable on artificial intelligence (AI) and machine learning (ML) as tools used to tackle issues regarding demand for foodstuffs, climate variance, and resource efficiency. This study has introduced an integrated Artificial Intelligence (AI) framework by integrating IoT based sensor data, Machine Learning (ML) models and deep learning algorithm for better crop monitoring and better crop decision making. IoT Data Capture of soil moisture, temperature, humidity, pH, light intensity and water TDS were used to train a Random Forest Regressor for predicting soil moisture, and a Random Forest classifier for detecting plant stress. A convolutional neural network also was developed, based on 54,303 RGB leaf images from PlantVillage dataset, to diagnose 38 classes of crop diseases and also historical USDA yield data was analyzed to provide a context to the long-term trends in productivity. The regression model showed a MAE of 5.07, RMSE of 14.80 and the stress classifier showed 95% accuracy, which exhibits great potential for real-time irrigation and real-time stress monitoring. The CNN received a validation accuracy rate of 90.45% which confirms its suitability for automated detection of diseases. Overall, the outcome of this study suggests that combining IoT sensing, ML prediction and DL based diagnostics is a promising way towards making Scalable, Intelligent Precision Agriculture Systems. Keywords: artificial intelligence, machine learning, precision agriculture, IoT, Disease detection.