Automated Identification and Classification of Rice Plant Leaf Diseases by Combining CNN and SVM

Rice serves as a fundamental staple crop on a global scale; however, its productivity is frequently jeopardized by an array of leaf diseases, which can result in substantial losses if not identified promptly. Conventional methods for disease identification are characterized by their labour-intensive nature, protracted timelines, and susceptibility to inaccuracies. This study introduces a machine learning-oriented methodology for the automated identification and categorization of rice leaf diseases, with the objective of enhancing the efficiency of disease management practices within the agricultural sector. The proposed system employs high-resolution imagery of rice foliage, which undergoes processing through various image preprocessing techniques, including resizing, normalization, and augmentation. Following this processing, the enhanced images are subsequently input into machine learning models for analytical purposes. VGG-19 is a subset of deep learning methodologies, are utilized owing to their exceptional capability for feature extraction from image datasets. The model is trained on an extensive dataset comprising labelled images of both healthy and afflicted rice leaves, encompassing diseases such as leaf blast, bacterial blight, and brown spot. A range of machine learning algorithms, including CNNs, SVM, and decision trees, are systematically assessed for their efficacy.