Isolated Gujarati Handwritten Character Recognition (HCR) using Deep Learning (LSTM)
Bhargav Rajyagor, Rajnish Rakholia · 2021 Fourth International Conference on Electrical, Computer and Communication Technologies (ICECCT) · 2021
With the swiftly escalating the paperless and automated offices and governance we required to convert paper into digital form. HCR is the form of optical character recognition to recognition the printed or handwritten text into digital text. The writing style of a person, size and thickness of the characters are different from person to person hence HCR is more challenging for automated system. As concern with Gujarati HCR there is more requirement to develop such an automated system that can be used for Gujarati Character Recognition with accents. [1]. In this study author have focused to develop LSTM model for offline Gujarati Character Recognition. Along with this an attempt to improve the rate of Gujarati character recognition using the LSTM model with the help of teaching and learning process of model with the given dataset of almost 58,000 images. The novelty of this proposed system is to identify a complete set of Guajarati characters that are available with the Unicode dataset. Author have used LSTM model in this study and achieved ~ 97% success rate of each character class.