Automated Polyglot Script Recognition and Translation Engine

In a linguistically diverse country like India, there is an increasingly urgent need for seamless multilingual communication. This project provides an Offline Multilingual OCR and Translation System allowing users to extract text from images, and convert or translate to a text output, and audio output, into a range of regional Indian languages, using Optical Character Recognition (OCR) dedicated to Tesseract, and a Neural Machine Translation (NMT) powered by Facebook’s NLLB-200 model, a language translation model trained on a large parallel corpus across languages like Kannada, Tamil, Telugu, Hindi, Marathi, and English. The application programmed for this system was made possible with the use of Flask, while a MySQL database engine is reponsible for securely authenticating users. The application allows users to upload an image, extract the text component, translate that text to another language, listen to the translated output, etc. The system has offline capabilities for when data privacy is a key consideration of the user, and to ensure users can access its functionality in real time and without needing an internet connection. However, using this solution will empower the real-world problems for education, the service sector for tourism, and public service delivery by providing an easy-to-use solution for anything regional in the translation domain.