Implementasi Metode Convolutional Neural Network (CNN) Dalam Klasifikasi Motif Batik.

Abstract

Indonesia is renowned for its diverse ethnicities, each contributing to a culturally rich mosaic. Among the distinctive regional traits, batik stands out prominently, evolving uniquely in each part of the country. However, the diversity in batik designs often confuses people trying to identify the region of origin due to similarities in patterns. Deciphering these unique batik motifs typically requires specialized knowledge, particularly from individuals well-versed in the art of batik. Reviews suggest that employing pattern recognition methods is an effective way to tackle this challenge. In today's technological landscape, various methods have emerged to aid in recognizing fabric motifs. This study utilizes the Convolutional Neural Network (CNN) method with the Efficient Net-B0 architecture. The tests conducted to identify batik motifs using this approach yielded a highest accuracy result of 79.62% for the test data and an accuracy validation result of 73.33%. These findings underscore the potential of advanced technologies, specifically the CNN with Efficient Net-B0 architecture, in accurately discerning and distinguishing batik motifs.