• Vadapalli Kamaswara Kiran
• Durga Ganga Rao Kola

Devanagari is one of the most widely spoken languages in India, an ancient nation with many varied languages and civilizations. It is used in several other Asian nations in addition to India. Devanagari manuscripts provide a wealth of knowledge that can be used to advance understanding in a variety of subjects. Due to their age, writing style, ink stains, and damaged pages, these manuscripts are decaying. These manuscripts can provide information when they are digitalized. In order to recognize the characters in Devanagari manuscripts with better optimization, optical character recognition (OCR) is vital. Convolutional neural network, a potent technique in the field of pattern recognition, is utilised to classify these identified characters. Convolutional neural networks are used in this paper as a feature extractor and classifier for identifying these documents. 46 classes of Devanagari characters are taken into account, testing and validation data are separated. With the suggested CNN architecture, recognition accuracy of 98.26% is achieved using 75% training data and 25% testing data.

Devanagari manuscripts, Optical Character Recognition, Convolutional Neural Networks, Feature extraction, Classification.

"A NOVEL DEEP LEARNING ARCHITECTURE FOR DEVANAGARI CHARACTER RECOGNITION", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.9, Issue 10, page no.d127-d133, October-2022, Available :http://www.jetir.org/papers/JETIR2210328.pdf

"A NOVEL DEEP LEARNING ARCHITECTURE FOR DEVANAGARI CHARACTER RECOGNITION", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.9, Issue 10, page no. ppd127-d133, October-2022, Available at : http://www.jetir.org/papers/JETIR2210328.pdf