Implementasi Metode Convolutional Neural Network Pada Pengenalan Aksara Bali Berbasis Game Edukasi

Abstract

Balinese script or also known as hanacaraka, is the writing used by Balinese people to write their language. In general, this script is used to write everyday language and literary language. Balinese script in the past was not only used for writing literature or sacred texts but also for writing everyday language. Balinese script plays an important role in literary writing. The sacred text of the Vedas uses Balinese script in the Sanskrit language. In preserving the Balinese script, itself, Balinese script lessons are mandatory for students from elementary school to high school. In addition to studying at school, interesting learning is certainly needed to attract students' interest. One way is by way of game applications or educational games. This Balinese script recognition application receives input in the form of Balinese script writing characters from the user, then it will be processed by preprocessing and continued with the classification training process using the Convolutional Neural Network (CNN) and Backpropagation methods. The result is a web-based application that can recognize Balinese script writing with the CNN classification method with an accuracy rate of 81.3% and gets a positive response from respondents who have tested the application.