Design and Development of Character Recognition Applications on Mobile Devices for Practice writing Thai consonants

Abstract

This research aims to: 1) Design and develop character recognition applications on mobile devices for practice writing Thai consonants 2) Study the satisfaction of 30 users of the application. This research process is as follows: 1) Data preparation 2) Creating a model for pattern recognition 3) Application development 4) Using the application 5) Evaluation. We collect handwritten data on a tablet from 100 computer science students. All images in datasets contain 4400 from 44 characters, divided into two parts 1) training data 3080 images and 2) testing data 1320 images. The datasets are processed by image processing techniques for suitability to use in the next step. Character recognition models were created by Teachable Machine tool. Android Studio and TensorFlow Lite use to create the application. The evaluation accuracy of character recognition models is 94%, and user satisfaction assessment results of the application at a good level (􀝔􀒧 = 4.07, S.D. = 0.80), Respectively.