Designing Batik and Artificial Batik Differentiator Applications Using Tensorflow

Autor: Isnaini, Dwi Wiji Lestari, Paras Trapsiladi, Zohanto Widyatmoko, Euis Laela, Irfa'ina Rohana Salma, Masiswo, Setiawan, Joni, Vivin Atika, Yudi Satria, Agus Haerudin, Guring Briegel Mandegani, Arta, Tri Kusuma, Novita Ekarini, Tika Sulistyaningsih, Syabana, Dana Kurnia
Jazyk: angličtina
Rok vydání: 2019
Předmět:
DOI: 10.5281/zenodo.3470830
Popis: Batik is the pride and masterpiece heritage of Indonesia. Batik has awarded as cultural heritage from UNESCO on October 2nd, 2009 and it is significantly affected to batik industry afterward. UNESCO has recognized batik as a traditional textile produced by using a technique of wax-resist dyeing applied to the fabric. Many products are produced to resemble batik, however along with technological innovations the products do not use hot wax. It will be difficult for the public to distinguish between the real batik and artificial batik available on the market. In order to determine whether the item is a real batik product or a falsified product, a tool is required. Center for Crafts and Batik has conducted research on making the "Batik Analyzer" software. The software's design is made using "deep learning" technology. The software uses TensorFlow. TensorFlow is a computational framework for making "machine learning" models. Machines are trained to be able to learn the difference in products by providing learning tools in the form of authentic batik which are: stamp batik and hand-written batik; artificial product which are: batik print imitation, cold wax batik print imitation, and burn-out print batik imitation products. The application that was created later was designed to run on an Android-based smartphone or tablet. Applications are then trained to recognize the product. The used data set is taken at The Center for Crafts and Batik for each type of fabrics to be distinguished. The results of the exercise are tested to be able to recognize the product. From the results of the training, it can be found that the software can distinguish products with 70 percent accuracy.  
Databáze: OpenAIRE