Hand Written Text Recognition using Convolutional Neural Network and Recurrent Neural Network

Character recognition is one of the developing research fields in Machine Learning. Most of the ancient famous literatures written by many famous writers are in the form of hand written text. Most challenging work of researcher is to extract these hand written texts and store digitally. In hand written character recognition, the characters are extracted and computers can be able to read and edit texts. The character recognition improves the manmachine interface for various applications which in turn helps for the development of fully automated system. Traditional approach of and written character recognition involves Support Vector Machine (SVM) classifiers to differentiate between the writers and some of the features like shape of the letters, spacing between the letters were extracted. In this paper Convolutional Neural Network and Recurrent Neural Network (CNN-RNN) model is designed to identify the features. CNN-RNN model is a deep learning based approach which uses a small patch of hand written images. The feature is extracted in CNN-RNN model and trained with a generalized binary form of logistic regression classifier.