Accuracy Evaluation of Plant Leaf Disease Detection and Classification Using GLCM and Multiclass SVM Classifier

Abstract

Plants are extremely disposed to diseases that mark the growth of the plant which in chance marks the natural balance of the agriculturalist. The yield of crop drops due to contagions instigated by numerous types of illnesses on parts of the houseplant. Leaf illnesses are principally instigated by fungi, bacteria, virus, etc. Verdict of the illness would be completed precisely and suitable activities should be occupied at the suitable period. Image processing techniques are more significant in detection and classification of plant leaf disease, machine learning models, principal component analysis, probabilistic neural networks, fuzzy logic, etc. Proposed work designates in what way to notice plant leaf illnesses. The proposed scheme will deliver a profligate, natural, accurate, and actual reasonable technique in identifying and categorizing plant leaf diseases. The proposal method is intended to support in the identifying and categorizing plant leaf illnesses using multiclass support vector machine (SVM) classification method. First, input image acquisition, second, preprocessing of images, third, segmentation for discovering the affected region from the leaf images by utilizing K-means clustering algorithm and then gray-level co-occurrence matrix (GLCM) features like (color, shape, and texture) are mined for classification. Finally, classification technique is utilized in identifying the category of plant leaf disease. The projected method compared with K-nearest neighbors (KNN) algorithm and successfully identifies the disease and also classifies the plant leaf illness with 96.65% accuracy.