A Comprehensive Survey on Handwritten Gujarati Character and Its Modifier Recognition Methods

Abstract

In India, handwritten character recognition is becoming necessity regional wise due to new education policy 2020. Various technologies are applied to solve the problem in this area like statistical or probability model, support vector machine, Bayes probability model, deterministic finite automaton (DFA), hidden Markov model, and many more which are used. Due to the advancement in machine learning, convolutional neural network is a good solution of HCR which gives more promising results but any new algorithm in machine learning that depends on training data, mathematical function, loss function, and method of evaluation of model. Focusing on past research of handwritten Gujarati character recognition is found that sufficient research is required for modifier level called “Barakshari”. Results obtained in past are limited to character level only. In this paper, our effort is to analyze and summarize previous contributions in the handwritten character recognition for several Indian languages