ULOS FABRIC CLASSIFICATION USING ANDROID-BASED CONVOLUTIONAL NEURAL NETWORK

摘要

Indonesia is a country with diverse ethnic, religious, and cultural backgrounds. Among the tribes in Indonesia, one of them is the Batak tribe. The Batak tribe has a variety of cultures, one of which is ulos fabric. Every ulos fabric has meaning, and its patterns also have different meanings. However, in today's world, ulos fabric has begun to be forgotten. Many of ulos models and types are circulating, yet people can barely recognize them and sometimes they even do not know that it is a ulos fabric pattern. So to preserve the culture of this ulos fabric, we tried to classify the ulos fabric using the convolutional neural network method. We selected Convolutional Neural Network (CNN) because it shows better results in image recognition in recent years. We get the accuracy of around 87.27% in different factors. The model is then deployed to Application Programming Interface (API) to be used in android application that can predict the ulos fabric. The aim of the application and research is to help the people to recognize the ulos fabric pattern by taking pictures of it and then they will get information about the function of the ulos fabrics and its history that lies behind it.