부산대학교 도서관

The devastating crop losses and economic damage caused by mango leaf spot disease pose a danger to the global mango industry. To classify mango leaf spot disease seriousness across six categories, we present a hybrid deep learning (DL) model that combines Convolutional Neural Network (CNN) and Support Vector Machine (SVM). The model is trained and tested using a dataset of 20,000 photos of diseased mango leaves gathered from a variety of sources, including internet repositories and field surveys in mango plantations. In the suggested hybrid model, a CNN is utilized for the feature extraction process, while an SVM is utilized for the classification process. High-level features from the input pictures are retrieved by the CNN component, and the SVM component is then trained to classify the images based on those features. On the assessment set, the proposed model demonstrated its generalizability and robustness by attaining an overall accuracy of 95.68%. These findings validate the efficacy of the proposed model in detecting and categorizing photos of mango leaves affected by various diseases. High accuracy was also achieved on an independent dataset, demonstrating the model's generalizability and resilience. The suggested approach may help farmers and agronomists detect and track the seriousness of mango leaf spot disease, which would lessen crop losses and boost crop output and quality. This research makes a difference in agriculture and plant pathology by giving a reliable method for keeping mango crops healthy. The suggested hybrid DL model provides a framework for creating dependable and practical tools for disease diagnosis and management in agriculture. It shows promise for the multi-classification of mango leaf spot disease based on six degrees of seriousness.