Rice Leaf Disease Detection and Remedies using Deep Learning

Abstract

Rice, a critical food staple cultivated globally, faces significant productivity challenges from leaf diseases that reduce yields and cause economic damage. Traditional disease identification methods are typically manual, slow, and require specialized expertise. Automated detection systems offer farmers a way to identify leaf diseases with minimal human involvement. Earlier research in rice leaf disease identification primarily employed image processing techniques combined with machine learning algorithms. These approaches extracted specific features such as color attributes, textural patterns, vein structures, and lesion morphology from images of diseased leaves. Machine learning models then classified diseases based on these extracted characteristics. Deep learning approaches have emerged as particularly effective for disease detection, as they can identify complex patterns from large datasets without requiring explicit feature extraction. This systematic review examines various deep learning methodologies applied to rice leaf disease detection, including transfer learning, ensemble approaches, and hybrid techniques. The review also evaluates how effectively these methods address various challenges in the field. The review provides detailed information on model architectures, hyperparameter configurations, fine-tuning strategies, and performance evaluation metrics used across different studies. Additionally, it addresses limitations in existing research and suggests potential directions for developing more accurate and efficient rice leaf disease detection systems in the future.