Classification Of Rice Leaf Disease Through Image Analysis And Deep Learning

Abstract

This work presents a comprehensive deep learning method for precise and effective disease prediction of rice leaves. With Convolutional Neural Networks (CNN), I developed a model that outperforms conventional methods for diagnosing illnesses of rice leaves. Images of both healthy and injured rice leaves were added to the dataset after thorough preprocessing to guarantee model robustness. Appropriate hyperparameters found by an ablation study led to the successful deployment of a CNN model. My model achieves 98.35% training accuracy and 95.38% and 95.22%, respectively, sensitivity and precision rates. The model was able to distinguish healthy leaves with reliability as seen by the high Specificity and Negative Predictive Value (NPV) and low False Positive Rates (FPR). Predictive dependability of the model is shown by the Matthews Correlation Coefficient (MCC) of 94.67%, and precision-recall balance by an F1 score of 95.20%. Because accurate and early disease identification can significantly affect crop yield and food security, this research has broad societal implications. Environmentally speaking, the strategy promotes environmentally friendly farming methods by enabling focused actions that can lower pesticide use. Ethically speaking, the study addresses data privacy and equal access to technology. Future study is discussed on the possible extension of the model to other crops and its connection with Internet of Things devices for real-time monitoring. This work not only progresses agricultural AI but also sets the standard for next studies on environmentally friendly farming methods.