A CNN and Random Forest Fusion with Optimized Convolutional Layers for Accurate Disease Identification in Rice Using Machine Learning

This study introduces a unified method for illness detection in rice by integrating Convolutional Neural Networks (CNNs) and Random Forest. Brown Spot detection (accuracy of 93.19%) and False Smut recall (89.80%) stand out as particular points of strength for the model, which was trained on a dataset that also included Bacterial Blight, Sheath Blight, False Smut, Stem Rot, and Brown Spot. Table 3 outlines the architecture, which features a combination of three layers of convolution and later integration with Random Forest to strike a good balance between complexity and interpretability. The micro-average accuracy of 83.25 percent demonstrates the reliability of the model as a whole. While the study shows promise, it also points to areas where refining is needed to better deal with certain types of misclassification. The model's comprehensibility and technological advances set it apart as a useful tool in precision agriculture, providing farmers with useful insights that contribute to food security on a global scale. Proactive methods for disease control could benefit from field deployments and continuous monitoring in the future.