Detection of Potato Leaf Disease Using Multi-Class Support Vector Machine Based on Texture, Color, and Shape Features

Potatoes are one of the most widely grown foodstuffs in Indonesia. Potato cultivation requires special attention, especially to the symptoms of the disease. The approach often used to identify plant diseases is the direct observation of symptoms by experts. However, this approach is not efficient because the identification process takes a long time. Therefore, in this study, a system for potato leaf disease detection was developed using image segmentation and a Multi-Class Support Vector Machine. This study begins with leaf spot extraction using the Grabcut method and multi-Otsu thresholding. The results of image segmentation are extracted based on color characteristics using a Red Green Blue (RGB) color histogram, texture features using a Gray Level Co-occurrence Matrix (GLCM), and shape features using calculations of leaf spot contours. Then the classification process is carried out using the Multi-Class Support Vector Machine (SVM) with a linear kernel. This study produces a system that can detect potato leaf disease with an accuracy of 97.56%.