OCR using CRNN: A Deep Learning Approach for Text Recognition

Abstract

Optical Character Recognition (OCR) is a widely used technology that converts image text or handwritten text into digital form. However, recognizing handwritten text, printed text, and image text poses a significant challenge due to variations in writing styles and the complexity of characters. This paper proposes a novel approach for OCR using Convolutional Recurrent Neural Network (CRNN) that combines convolutional neural networks (CNNs) and recurrent neural networks (RNNs). The proposed CRNN architecture can automatically learn and extract features from raw image pixels and recognize sequential patterns of characters. This research paper presents a robust OCR system using CRNN architecture with 7 convolutional layers and 2 LSTM layers for recognizing text in images with complex backgrounds and varying fonts. The proposed system achieved state-of-the-art performance on several benchmark datasets, demonstrating the effectiveness of the proposed approach. Our experimental results demonstrate that the proposed CRNN approach is better than other methods and achieves higher accuracy with less latency in recognizing text from an image. We also analyze the impact of different parameters, such as the number of layers, filter sizes, and hidden units, on the performance of the CRNN model. This paper provides a comprehensive study on OCR using CRNN and its potential to improve the accuracy and efficiency of recognizing text.