Indonesian Traditional Shadow Puppet Image Classification: A Deep Learning Approach

Abstract

Puppet is one of Indonesian traditional art which is an art performance performed by a puppeteer. The puppets have their own features and character according to the puppet character. So one way to preserve it, we do research to recognize and classify the type of puppets. So with this can be known the types of puppet based on shades and shapes. With the differences of these characteristics, the writer made the classification of puppets by using Deep Learning method which is currently popular on computer vision. In this study we used Deep Convolutional Neural Network architecture such as AlexNet and VGG-16. By using our own dataset, we classify the types of puppets based on the puppet feature with the number of databases that we have after going through the data augmentation is 5130. The best result we made is using architecture VGG-16.