Gray Level Co-Occurrence Matrix Algorithm and Backpropagation Neural Networks for Herbal Plants Identification

Abstract

Leaves are one of the most crucial plant structures. Leaves function as absorption, processing food through photosynthesis, and as a means of transpiration. Various medicinal plants have been known to the public for a long time. However, it is still difficult for ordinary people to remember that all leaves are relatively the same color, green, and do not know the characteristics of the leaves. In this paper, we will detect the types of herbal leaves of medicinal plants by identifying leaves using the GLCM (Gray Level Co-occurrence Matrix) algorithm and artificial neural networks backpropagation. The dataset obtained from the Kaggle-Leafsnap dataset has as many as 50 types of herbal plant leaves. The GLCM algorithm is a method with probability and statistics. This method extracts leaf images, converts the image into grayscale, and then changes the shape of the image data into numerical. The GLCM algorithm is combined with Backpropagation in classifying data. There are several stages in the classification process in this study, namely data retrieval, data preprocessing, data classification, and evaluation method. This research aims to achieve high accuracy through the proposed method.