Pengembangan Sistem Pengenalan Karakter Aksara Suku Simalungun Berbasis Android

Abstract

Simalungun script is the script used by the Simalungun people to communicate with each other in their time. But over time, this character is rarely used. Therefore, the Simalungun script needs special attention because it is already threatened with extinction due to limited data and information. To overcome this, it is necessary to use the role of information technology. In this study, a system was built that can classify and introduce the Simalungun tribal characters using the Android-based Convolutional Neural Network (CNN) method. In addition, in this study, CNN MobileNetV2 and TensorFlow Lite architectures were used for deploying android needs. From the results of the training using the MobileNetV2 architecture by testing 29 characters, the accuracy results are 81% and the test results are 82%. In testing the feasibility of the application, the author uses the concept of usability testing by involving 15 respondents and giving a percentage of 78%.