PENGENALAN MOTIF BATIK PESISIR PULAU JAWA MENGGUNAKAN CONVOLUTIONAL NEURAL NETWORK

Abstract

Coastal Batik is made outside of Solo and Yogyakarta. The use of the term "coastal" is due to the majority of batik production being located in the northern coast of Java, such as Indramayu, Cirebon, Pekalongan, Lasem, and others. Coastal batik is characterized by flexible color selection and patterns, influenced by foreign influences, particularly after the introduction of Islam in the 16th century. The Convolutional Neural Network (CNN) method is commonly used in classifying digital image data. Neurons in CNN are represented in a two-dimensional form, differing in linear function and weight parameters. The CNN extraction process consists of hidden layers, including convolutional, pooling, and ReLU (activation function) layers. The evaluation results of the Convolutional Neural Network model show that it can perform classification and recognize coastal batik images of Java Island, achieving the best results in the first scenario with a training data ratio of 70% and testing data ratio of 30%, resulting in an accuracy of 83%. For future research, it is recommended to increase the number of batik images and capture them directly, while incorporating segmentation or extraction features to measure efficiency and accuracy levels. This will help obtain better results in recognizing the characteristics of coastal batik in Java Island.