Identification of Batik in Central Java using Transfer Learning Method

Abstract

Identification of Batik in Central Java using Transfer Learning Method. Batik was recognized as a human heritage for oral and nonmaterial culture by UNESCO due to its symbolic and philosophical ties to the lives of Indonesians. However, the younger generation is gradually losing itslegacy because of technological and sociological changes that have influenced Indonesian batik. Consequently, batik knowledge is disappearing. A convolutional neural network and transfer learning techniques were utilized in deep learning to construct a model recognising batik motifs. The study utilized a dataset of one thousand images, five classes of batik designs (Banji, Kawung, Slope, Parang, and Slobog), and pre-trained architectural models VGG16 and VGG19 on Keras. The best model utilizes the VGG16 architecture, and the number of epochs is 50,with the result of testing accuracy of 0.9200.