Devanagari Character Recognition using Deep Convolution Neural Networks

In the presented paper, a Deep Learning architecture for Devanagari handwritten character recognition is introduced. The dataset comprises of 94742 images of 58 distinct classes containing numerals, vowels and consonants from Devanagari script. Deep Convolution Neural Network (DCNN) have indicated better outcomes than traditional Machine Learning algorithms and Shallow networks in recognition tasks. The proposed algorithm achieved an accuracy of 99.20% on testing data with 98.31% accuracy on training data.