Holistic Approach Employing Different Optimizers for the Recognition of District Names Using CNN Model

Abstract

A Holistic approach is proposed for the recognition of Handwritten district names of Punjab state which are written in Gurmukhi Script. For the purpose of recognition, a Convolutional Neural Network(CNN) using deep learning is employed.  Initially, the dataset of 22000 of images is prepared for all the 22 district names of Punjab state and later a CNNis employed. The proposed  CNN architecture having 12 layers is developed and employed using three optimizers: Adam, SGD and RMSprop for the recognition task. Best Average Validation Accuracy achieved for the proposed CNN architecture is 95% and maximum achieved validation accuracy is 99% achieved by employing Adam Optimizer.