One-Syllable Hangeul Script Handwriting Detection System Using Convolutional Neural Network (CNN) Method Based on Flask Framework

Sistem Deteksi Tulisan Tangan Aksara Hangeul Satu Silabel Menggunakan Metode Convolutional Neural Network (CNN) Berbasis Framework Flask

Abstract

Since the emergence of the Korean Wave phenomenon, many people have started to learn the Korean language. To learn the Korean language, an understanding of its writing system is required. Therefore, a system that can accurately and easily detect handwritten Hangeul characters is needed. One of the deep learning methods used for image processing is the Convolutional Neural Network (CNN), which is developed from a multilayer perceptron (MLP). The CNN model is trained using a dataset of Korean language written in Hangeul characters, comprising 2460 image data divided into 24 categories. The designed and implemented system in this research achieves a validation accuracy of 99.4% and a training accuracy of 95.5%. The trained model is also deployed as a website application to facilitate testing and can be used for interactive learning. Through the Flask-based website application, the prediction of Hangeul characters is tested ten times for each class category, resulting in an average accuracy rate of 95.41%.