Pengembangan Aplikasi Pengenalan Tulisan Tangan Abjad dan Angka Berbasis Convolutional Neural Network

Abstract

This research aims to make an application that recognizes and predicts handwritings of alphabets and numbers using Convolutional Neural Network (CNN). This application uses an incremental model with 2 steps for its development. The data used is EMNIST dataset, images of handwritten letters consists of Roman capital letters, Roman small letters, and Arabic numerals (0-9) that are split into 47 different classes. The model and application successfully predicted handwritings of alphabets and numbers with an average precision percentage of 76,24%.