Text localisation from natural scenes

• Main Author: Ng, Alphaeus Yue Jie
• Other Authors: Loke Yuan Ren
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science; Text localisation; Language translation; Single-page-application; React; Flask
• Online Access: https://hdl.handle.net/10356/175335
• Institution: Nanyang Technological University

This project aimed to create an open-sourced application that mass-translates natural scene images. The application consists of three main components: 1) Integration of open-source initiative, EasyOCR, for text detection, 2) Utilisation of cutting-edge language translation models, with a focus on the capabilities provided by OpenAI’s models, 3) A single-page application built on Flask and the React framework for accessibility and a user-friendly experience. By integrating these technologies, the project hopes to significantly advance the seamless and efficient localisation and translation of text, contributing to the larger landscape of natural language processing advancements. In addition to these solutions, the report carefully examines the rationale behind the components contributing to the overall system architecture.