Classification of Healthy and Diseased Mulberry Leaves using Machine Learning

Sericulture, an essential agricultural sector in India, depends greatly on the quality of mulberry leaves, serving as the primary nutrition source for silkworms crucial for silk protein synthesis. This study proposes a methodology utilizing K-Nearest Neighbor (KNN) and Naive Bayes (NB) algorithms with VGG-16 to classify healthy and diseased mulberry leaves. Both KNN and NB models utilize auto features extracted by VGG-16 during training to classify healthy and diseased mulberry leaves. NB models achieve slightly lower recall and precision rates of 89% and 93%, respectively, with a 91% F1-score, while KNN models achieve 90% recall, 100% precision and 95% F1-score for diseased leaves. In the healthy category, NB exhibits 89% precision, 93% recall, and 91% F1-score, whereas KNN showcases 91% precision, 100% recall, and 95% F1-score. On average, NB models achieve 91% accuracy, whereas KNN models exhibit a 95% accuracy. This comprehensive approach offers the potential for effective disease management, ensuring consistent high-quality silk production in sericulture.