KLASIFIKASI MOTIF BATIK INDONESIA MENGGUNAKAN CONVOLUTIONAL NEURAL NETWORK (CNN)

Abstract

 Batik is one of Indonesia’s cultural heritages, known for its diverse motifs, patterns, and colors, and has been recognized by UNESCO as an intangible cultural heritage of humanity. Despite this recognition, efforts to promote and preserve batik, particularly among younger generations and the global community, continue to face significant challenges. This study aims to develop a web-based Indonesian batik motif classification system utilizing machine learning technology, specifically Convolutional Neural Networks (CNN) with the EfficientNetB0 architecture. The system is designed to identify and classify 20 batik motif classes based on visual characteristics such as patterns, colors, and shapes. Through this interactive web-based platform, users can explore and learn about different types of batik, thereby fostering greater understanding and appreciation of this cultural heritage. The training process demonstrated notable improvements in accuracy, with the final model achieving a test accuracy of 82.5% and an average precision, recall, and F1-score of approximately 0.83. These results indicate that the developed system holds substantial potential for digital batik preservation and promotion, serving as an effective educational tool for a wider audience in the era of information technology.