Citrus Plant Leaves Disease Detection Using CNN and LVQ Algorithm

This study introduces a unique method for disease identification in citrus plants by combining convolutional neural network (CNN) and learning vector quantization (LVQ) techniques. The suggested technology is meant to aid in the early identification and diagnosis of citrus plant diseases, which is important for preserving crop yields and avoiding crop loss. Features are extracted from pictures of citrus plant leaves using a convolutional neural network. Furthermore, the LVQ algorithm is used to identify the retrieved features as either healthy or unhealthy. When tested on a dataset consisting of photographs of citrus plant leaves, the suggested system achieved a high accuracy of 96.33% in disease classification. A total of 3570 pictures were used in this analysis, including both healthy and diseased citrus plants, representing different pathogens (citrus canker, citrus scab, citrus rust, other diseases, and healthy images) classes. There are 500 test photographs from each class and 1070 full-size images throughout test categories. An F1-score of 96.54%, a recall score of 96.54%, and a precision score of 96.69% were all obtained using the proposed strategy. Based on the obtained data, it appears that the proposed method achieves superior accuracy in disease identification compared to the state-of-the-art methods. The citrus industry stands to benefit greatly from this strategy, as it may be used for early disease identification and prevention in citrus plants.