Improved Real time Signboard Detection and Translation Using FRCNN

Abstract

In India, majority of people are speaking the language Hindi, but a major portion of signboards are in English. On a business or pleasure trip, the travelers get confused by the various sign boards written in English. As smartphones becomes most popular inrecent years,they canrely on smartphone for the same. This paper explains the work intended to build a mobile application that can recognize the English content and sign present on the signboard image, detect and translate the content and symbols from English to Hindi and display the translated Hindi text back to the screen of the phone. The system uses pre-trained faster regional convolutional neural networkusing pre-trained CNN for object detection, tesseract OCR for text extraction and English-to-Hindi dictionary for translation.