ANALISIS DAN PERBANDINGAN METODE CNN DAN SVM DALAM MENDETEKSI BATIK NUSANTARA

Abstract

Abstract

Batik, as part of Indonesia's cultural heritage, possesses unique patterns specific to each region. This study aims to

automatically classify batik motifs from various regions such as Aceh, Bali, Betawi, Dayak, Papua, and others,

each with its distinctive characteristics, using Convolutional Neural Network (CNN) and Support Vector Machine

(SVM) methods. By comparing the performance of these two methods, this research seeks to identify the most

effective approach for distinguishing the visual characteristics of different batik types. The results of the study

show that the SVM method outperforms CNN in a scenario with only 20 batik categories and 100 images per

category. SVM proved to be more efficient in recognizing visual patterns with a limited amount of data, while

CNN requires more data to achieve optimal results. These findings have potential applications in batik

authentication, intellectual property protection, and the digital promotion of batik culture. This research evaluates

the effectiveness of both methods in recognizing batik visual patterns and demonstrates that SVM performs better

in situations with a more limited dataset.