Handwritten Character Recognition using Deep Neural Networks

Abstract

The preliminary work performed in this manuscript is to recognize Handwritten English Characters using a multilayer perceptron. The standard EMNIST dataset of handwritten English characters is used here. The preprocessing of images included Binarization and reshaping into 784 (28x28) binary pixels. The model was trained using a dataset of 80,000 characters. The validation set had 20,000 characters other than those of the training set. The recognition accuracy on the Training set is 98.05%, and that on the validation set is 91.46%.