Plant Leaf Diseases Detection Using KNN Classifier

Abstract

In the agricultural area today, identifying Leaf Disease is very challenging. A significant portion of the economy is dependent on agricultural output, thus when a disease is incorrectly identified, there will be a significant loss in crop yield and the economic value of food. The cost of traditional procedures is quite low, but they need human labour to visually inspect the plant's leaf patterns and identify the disease, as well as more time and laborious effort from farmers. Software-based methods for detecting leaf illnesses involve a significant amount of work, expertise in leaf diseases, and longer processing times for the diagnosis. Based on the KNN Classifier, we have suggested image processing for identifying leaf disease. The procedure for identifying a disease includes image preprocessing, contrast enhancement, RGB conversion, feature extraction, segmentation, and classification using K-nearest neighbour algorithms. The quality of the splint samples is improved by using Histogram Equalization, which is initially applied to the samples of leaves after they have been shrunk to 256x256 pixels and image-preprocessed. The instructional characteristics of the splint samples are valued using a matrix called the Grey Level Co-occurrence Matrix. Using machine learning techniques akin to K-Nearest Neighbor, the characteristics are categorized (K- NN). K- NN is used to evaluate the proposed model's quality.