A Comparison and Evaluation of Handwritten Digit Recognition (HDR) Algorithms

The technique of digitizing handwritten digit pictures is known as handwritten digit recognition (HDR). The process of translating information from paper format to digital one is an exceedingly expensive one. HDR is an approach that may be utilized to remedy this problem. The main objective of this paper to design an effective algorithm that is capable of recognizing scanned and transmitted user input that contains handwritten numerals. The objectives of this study are two-fold: the first is to evaluate the accuracy of several distinct algorithms that are capable of classifying handwritten numerals by making use of a range of different numbers of hidden layers and epochs; and the second is to investigate the degree to which these distinct methods differ from one another. In this paper, an accurate digit recognition system has been developed which will produce the simplest output and the dataset provided by MNIST has been utilized for this paper. It also handles the challenges that are faced due to the variations in input received from various users.