Hieroglyphs Language Translator using deep learning techniques (Scriba)
Ragaa Moustafa, Farida Hesham, Samiha Hussein, Badr Amr, Samira Refaat, Nada Shorim, Taraggy M. Ghanim · 2022 2nd International Mobile, Intelligent, and Ubiquitous Computing Conference (MIUCC) · 2022
Egyptian hieroglyphs are symbols that are part of the ancient Egyptian writing system. Hieroglyphs do not have a fixed writing direction, and a single glyph can have multiple definitions, making them difficult to translate into spoken language. Deep learning techniques, such as EfficientNet, MobileNet, and ShuffleNet, have been tested on two hieroglyph datasets (with and without data augmentation). Current experiments revealed that MobileNet and EfficientNet produced the most accurate results Ranging from 99% to 100%. Scriba is a flutter-based mobile application that shall use one of the mentioned algorithms to provide a precise translation of hieroglyphs. This paper thoroughly discusses the proposed development approach, experiments that were carried out (and their results), and the implementation of the mobile application.