Image-Based Text Detection and Multilingual Translation Using Deep Learning

Text present in images often becomes difficult to read when the image is blurred, captured at a low resolution, or affected by poor lighting and low contrast. These challenges can reduce the accuracy of conventional optical character recognition (OCR) systems, particularly when the text needs to be recognized and translated into another language. To address this problem, this research presents an intelligent deep learning-based OCR system for blurry text detection, recognition, and English–Kannada translation from images. The proposed system allows users to upload an image containing printed or scene text. It identifies text regions, draws bounding boxes around the detected text, and uses an OCR recognition model to extract the corresponding words or sentences. Image preprocessing techniques, such as contrast enhancement, resizing, and noise reduction, can be applied to improve text readability before recognition. The recognized text is then processed to provide translations in both English and Kannada, helping users understand content written in either language. The system displays the recognized text and its translations alongside the processed image, making the results easier to interpret. The primary objective of this work is to improve the usability of text recognition for images containing blurry or difficult-to-read text while providing a simple bilingual translation interface. The proposed approach combines text detection, image enhancement, OCR, and machine translation within a unified workflow. Its performance can be evaluated using precision, recall, F1-score, and OCR accuracy against appropriately annotated test data. This system has potential applications in reading signboards, printed materials, notices, and other image-based text where English–Kannada language support is useful.