Identifikasi Senjata Tradisional Mandau Suku Dayak Menggunakan Metode Support Vector Machine

Abstract

Mandau is a sharp weapon like a machete that comes from the culture of the Dayak tribe in Kalimantan. Mandau itself has many types of carvings with different motifs in each typical Dayak tribe. The diversity of Dayak clumps and sub-tribes produces various types of Mandau, which are similar. Most Indonesians cannot distinguish between the existing variety of Mandau because of the information lack about  Mandau. The study aims to identify Mandau types by applying Support Vector Machine (SVM) methods according to shape and texture feature extraction of image Mandau. Process of feature extraction using eccentricity, metric, contrast, correlation, energy, and homogeneity parameters. Six parameters obtained from the image feature extraction were used for the classification process. There are four kind of Mandau that used, such as Mandau Benuaq, Mandau Kenyah, Mandau Mahah, and Mandau Tunjung, with each Mandau 24 images, and the total data is 96. The implementation of the SVM method in identifying the type of Mandau using Linear Kernel, Gaussian / RBF Kernel, and Polynomial Kernel. The results indicated that when SVMs were used to identify Mandau images in split data, 80% of the training and 20% of the test data had the highest average accuracy of 82%. Based on the type of Kernel at 80:20 data split, the identification of the Mandau image using the Polynomial Kernel has the highest accuracy rate of 95%.