ComiTranslate: empowering global readership through auto-translated comics and manga

Goh, Yee Fong (2024) ComiTranslate: empowering global readership through auto-translated comics and manga. Final Year Project, UTAR.

Preview

PDF Download (3623Kb) | Preview

Abstract

Due to the steep learning curve and time needed to translate the contents of the comic, readers are faced with significant challenges due to this language barrier. Conventional methods often include the laborious and time-consuming use of dictionaries or other language-learning aids, which can be discouraging. Therefore, this project aims to encourage readership of foreign literature by providing a solution which automates the translation process by the press of a button. The proposed system utilizes Python Image Library, Tesseract, Keras�OCR, Manga OCR, DeepL, and OpenCV to achieve the desired result as a solution. Kanban In Agile Methodology was practiced throughout the development phase as a way to manage the project’s progress. The application detects all computer generated text, extracts them, translates, and then places the translated text back onto the image . Due to the application’s superiority in its translation speed as compared to human translators and redrawers, it is able to make reading self-published works, which are often not translated by official sources, easier for readers as the application provides them the means of translating the comics. However, there are still flaws in its accuracy and improvements are planned for future iterations of this project. In conclusion, all objectives had been fulfilled through the development of this project.

Actions (login required)