Thresholding based Tomato Leaf Disease Classification

The tomato is a large consumable vegetable; whose production suffers due to various types of plant diseases at every stage of a plant’s growth under various circumstances. Out of all the diseases, mosaic, curl viruses, and leaf mold are the most dreadful tomato leaf diseases, which are recorded all over the world. The computerized detection of these diseases helps in developing an Automated Fast Detection and Decision Making System for the identification of various types of diseases on the Tomato Plant. This paper explains the application of a Kapur-based thresholding technique for the detection of mosaic, curl virus, and leaf mold infected regions on tomato leaves. And for the classification of these diseases, the features are extracted, and classification is performed using six different machine learning based classifiers: KNN, Ensemble Classifier, Discriminant Analysis, Decision Tree, SVM, and Naive Bayes for comparing results. The proposed method accurately detects the disease infected area and scores with a maximum of 90% accuracy through the KNN classifier, which outperforms other classifiers. So, while designing computerised diagnosis of mosaic, curl, and leaf mold viruses on tomato plant leaves, the proposed approach can be considered.