학술논문
A Regularized CNN-SVM Framework for Improved Diagnosis of Citrus Fruit Diseases, both Common and Exotic
Document Type
Conference
Author
Source
2023 International Conference on Sustainable Computing and Smart Systems (ICSCSS) Sustainable Computing and Smart Systems (ICSCSS), 2023 International Conference on. :407-411 Jun, 2023
Subject
Language
Abstract
The current study presents a deep learning method that has been developed to detect and categorize illnesses that impact citrus crops. The effectiveness of a convolutional neural network, a type of artificial intelligence, is enhanced by this approach. Architecture with the method involves using support vector machines to classify and then employ three sets of convolution and two completely connected layers, pooling layers, as well as other elements that make up the model. The dataset utilized in the samples produced images of citrus plants that have been afflicted by 9 distinct different kinds of diseases. To assess the model’s performance, various metrics such as precision, recall, F1-score, support, accuracy, and average are used. The effectiveness of the model is evaluated based on these metrics, and it was found to have an average weighted F1 score of 86.10% and an overall accuracy of 86.03%. The model achieved the highest precision score of 86.96% for the Citrus nematode class and the lowest precision score of 84.00% for the Dothiorella blight class.