Klasifikasi Jenis Kurma Berdasarkan Fitur Warna, Bentuk, dan Tekstur Dengan Support Vector Machine

Abstract

This study was conducted to address the difficulty in distinguishing between the widely distributed date varieties in Indonesia ie Ajwa, Sukari, and Medjool, which often share similar visual characteristics. The objective of the research was to develop an automatic classification system based on digital image processing and machine learning. Date fruit images were collected and processed through several stages, including preprocessing, segmentation using adaptive thresholding and morphological operations, followed by feature extraction covering color (HSV mean and standard deviation), shape (area and Hu moments), and texture (GLCM). The extracted features were then classified using a Support Vector Machine (SVM) with k-fold cross-validation and parameter optimization via Grid Search, employing both linear and RBF kernels. The results demonstrated very high performance, with an average accuracy of 96.86% for the linear kernel and 88.33% for the RBF kernel, with the best model obtained using the linear kernel at C=100. The minimal classification errors indicate that the extracted features were effective in consistently distinguishing the three date varieties. These findings confirm that classical feature-based methods combined with SVM remain highly competitive and efficient, particularly with limited datasets. The implications of this research highlight its potential application in supporting the imported fruit trade industry as well as serving as a reference for future studies in agro-informatics.