Enhancing Sunflower Disease Identification with CNN-SVM Integration

A worldwide crop, sunflowers are subject to illnesses that can reduce productivity and quality. Agriculture needs early and precise disease diagnosis to control and reduce economic losses. We present a new sunflower illness classification method using CNNs and SVMs. The study involves four primary phases. We initially collected sunflower leaf photos of healthy and diseased leaves, including Alternaria leaf blight, Downy mildew, Phoma blight, and Verticillium wilt. Resizing, contrast modification, and color improvement improve dataset quality and consistency. Phase two comprises data separation and feature extraction. The dataset has training, validation, and testing sets, and a pre-trained CNN model extracts key picture attributes. CNN penultimate layer features are used for categorization. Disease categorization neural network architecture is devised and executed in the third step. This design uses three convolutional layers for feature extraction, three max-pooling layers for dimension reduction, and one fully connected layer for classification. Optimized network structure extracts relevant features and makes accurate predictions. After extracting neural network features, Support Vector Machines (SVMs) are classified in the fourth step. Our model is assessed using accuracy, precision, recall, F1-score, and confusion matrices. On the same dataset, we compare our technique to others. The CNN-SVM hybrid technique for sunflower disease classification shows promise for precision agriculture and disease control. This research advances disease detection methods, giving sunflower producers a viable tool to improve crop health and production.