A Novel Text Recognition Using Deep Learning Technique

The branch of OCR (Optical Character Recognition) face many challenges such as extracting text from images. These challenges are most prominent due to varied nature of images in terms of size, letter style, orientation, background, quality, and format. This challenging problem requires modern solutions directed towards the proper and precise text extraction from different types of images. The proper handling of these problems is critical for the necessary increase of OCR systems' level efficiency. This paper tracks down the changes in the base assumptions of OCR due to the arrival of the deep learning algorithms. Convolutional Recurrent Neural Networks (CRNN) along with attention mechanism is suggested in this paper in order to enhance text recognition performance.