<![CDATA[Abstract : Tubers are plants that can grow in the yard, low-lying areas, or highlands and areas that lack water. But not all people in Indonesia plant tubers. So that some people do not recognize the types of tubers. The purpose of this study is to classify tuber types based on texture features, shape features, and both texture features, and image shapes. The benefits of this research are in order to provide information related to the types of tubers. The algorithm used for tuber type classification is SVM and KNN. The types of tubers classified as cassava or cassava, sweet potato and taro. The dataset used in this study were 180 images, with 150 images as training data, and 30 images as test data. The results of the trial show, the classification of tuber types using SVM algorithm with 10% accuracy texture features, with 7% accuracy form features, and both 7% accuracy features. While the tuber type classification uses the KNN algorithm with K = 5, the successive accuracy values for texture features, shape features, and both features are 3%, 20%, 7%. And if KNN with K = 2, the successive value of succession in texture features, shape features, and both features, is 13%, 23%, 10%. The shape features here are: area, perimeter, metric, major axis, minor axis, eccentricity. And texture features, namely: average intensity values of grayscale, standart deviation image grayscale, contrast values, energy, correlation, and homogeneity. Keywords: tuber types, SVM, KNN, texture and shape features]]>
Copyrights © 2020
