A novel deep learning model for cabbage leaf disease detection and classification

Abstract

Abstract Manual observation bias in identifying cabbage leaf diseases necessitates efficient automated detection systems in agriculture. In this study, a deep learning-based approach was developed for the automatic recognition and categorization of cabbage diseases, focusing on aphids, worm cuts, and black rot. Using 1400 images, including healthy and diseased samples, our method employs novel deep learning models. EfficientNetB7, MobileNetV2, and DenseNet201 were utilized to successfully identify three types of cabbage leaf diseases with accuracies of 98.56%, 98.26%, and 97.64%, respectively. EfficientNetB7 achieved the highest accuracy of 98.56% for disease classification. The system's ability to automatically extract features from images enhances disease detection, crucial for mitigating yield losses. Through visualization techniques, we analyzed the location of unhealthy regions in the leaves, the filter, and the intermediate layer.

Keywords

• Cabbage leaf disease
• Transfer learning
• Visualization