Klasifikasi Gambar Aksara Jawa Menggunakan Optimalisasi Parameter SVM dengan Kernel Cosine

Priyambodo, Aji
Prihati , Prihati

MEANS (Media Informasi Analisa dan Sistem)
Volume 9 Nomor 2

Pattern recognition of Javanese script (Hanacaraka) plays a vital role in cultural preservation through digital technology. This study aims to develop a classification model for Javanese script images using Support Vector Machine (SVM) with a Cosine similarity kernel, supported by parameter optimization to enhance accuracy. A dataset of 4,385 images underwent preprocessing and feature extraction using Histogram of Oriented Gradients (HOG). SVM parameter optimization via GridSearchCV resulted in a significant accuracy improvement. The proposed model achieved a 99.84% accuracy, outperforming previous methods such as CNN-SVM and DCNN. This study demonstrates the effectiveness of Cosine similarity in Javanese script recognition and contributes to the advancement of machine learning-based classification systems.