Handwritten Digits, English Characters and Te...

Handwritten Digits, English Characters and Telugu vowel Character Recognition using Convolutional Neural Network

Author Name : S S N L Priyanka, M Amith, B Ramya Bhagyalatha, G Nikhil, B Amulya Jyothi

ABSTRACT

Trending technologies like machine learning and Neural networks are used in different kinds of handwritten pattern recognition in various research areas. Hence, it is very hard to recognize different persons written digits, English alphabet characters with different stroke of writing. In this paper, we try to predict the images of Digits, English characters and Telugu vowel Characters from the dataset with the knowledge of Convolution Neural Network which is the most appropriate algorithm for classification of the images.

 It has been an important research area that recognizes the handwritten digits, English and Telugu characters. The English characters and digits are of the balanced dataset and the telugu vowel characters are the images of 6 different vowels of telugu characters.

Here an Artificial Neural Network model is trained and the accuracy of the model is 70 percent without the hidden layers. Then added the Dense layers as the hidden layers and trained the model and the accuracy obtained is 75 percent. As when we are trying to create a model with simple Artificial Neural Network with and without hidden layers the accuracy of the model is very low. So to solve this problem with better accuracy we trained a model with Convolutional Neural Networks which performs better in classification of the images.

Keywords: Artificial Neural Network, Convolution Neural Network, EMNIST, Telugu vowel dataset