Disease Classification of Tomato Plant Leaves Using Image Processing and Machine Learning Techniques
Waseem Ullah Khan, Secured IoT Devices Lab University of Engineering & Technology Peshawar, Pakistan.
Sohail, Secured IoT Devices Lab University of Engineering & Technology Peshawar, Pakistan.
Ilham Hamid, Secured IoT Devices Lab University of Engineering & Technology Peshawar, Pakistan.
Muhammad Nauman Khan, Secured IoT Devices Lab University of Engineering & Technology Peshawar, Pakistan.
Ebtihaj Abdullah, Secured IoT Devices Lab University of Engineering & Technology Peshawar, Pakistan.
Corresponding Author:
Waseem Ullah Khan (waseem@uetpeshawar.edu.pk)
Abstract:
Plants have an essential role in the cycle of nature. Only plants have the ability to convert the light energy emitting from the sun and then prepare their food. The reduction in production occurs due to plant diseases which is the main cause of economic losses. In plants, Tomatoes are famous for their vitamin content which contains beta-carotene (converts to vitamin A when consumed), vitamins C, vitamin E, vitamin B, and vitamin K, contributing some minerals like magnesium and calcium. The quality and production of tomatoes can be badly affected by diseases. In order to classify diseases in plants, techniques like computer vision and image processing have been used from the last decade. For identification and classification of Tomato diseases, this paper proposed a model which is trained on Tomato plant leaves. In the first step, the leaf regions are being identified by performing the image segmentation with the help of the image processing technique. Then, Histogram of Oriented Gradients (HOG) is used to extract features from the segmented images and train six different machine learning models (Support Vector Machine (SVM), Naive bayes, Decision tree, K-nearest neighbor (K-NN), Random Forest, Logistic Regression and Decision Tree) on Tomato plant leaves using HOG features. The proposed models are tested on the image dataset of five different classis (Tomato bacterial spot, Tomato healthy, Tomato mosaic virus, Tomato spotted spider mite, and Tomato yellow leaf curl virus). After comparison of the machine learning classifiers, SVM ranked first amongst the classifiers that achieves 92% classification accuracy on Tomato plant leaves disease image dataset.
Keywords:
Agriculture; Disease Detection; Image Processing; Image Segmentation; Image Features; Machine Learning