Gujarati handwritten character and modifiers recognition using deep hybrid classifier

Abstract

In the area of handwritten character recognition many researchers have worked and still working to achieve remarkable result. For the performance improvement of Indic and non-Indic scripts recognition, the necessary condition is to acquire proper domain knowledge and its intricacies otherwise research cannot be fruitful. Here, a Deep Hybrid Learning Classifier with fusion of convolutional neural network has been proposed that learns deep features for offline Gujarati handwritten character and modifier recognition (GHCMR). The proposed model works competently for training as well as testing and exhibits a good recognition performance. The datasets comprising huge image set of offline handwritten Gujarati characters with modifiers have been employed in the present work. The testing accuracies achieved using the proposed network is 97% for characters with modifiers.