Handwritten Text Recognition using Deep Learning Algorithms

Since a pen is more convenient than a keyboard, most scripts are now produced by hand; this often leads to mistakes due to the illegibility of human handwriting. To combat this issue, handwriting recognition has rapidly emerged as a top research priority. Computer vision algorithms involving optical character recognition were previously employed in traditional handwriting recognition systems. It is a challenging undertaking to train an optical character recognition (OCR) system with these constraints in mind. The OCR method has many problems. In this study, we employ Convolutional Neural Networks (CNNs), Long Short-Term Memories (LSTMs) built on Recurrent Neural Network (RNN) architecture, and Connectionist Temporal Classification (CTC) to recognise handwritten text (CTC). To train and evaluate the network, we use the Information Acquisition MNIST dataset, which includes an English language handwriting test. Here, image processing is handled by OpenCV, while word recognition and training are handled by TensorFlow. Python is used throughout the development of this system, with the console serving as the final destination for the output.