CNN Based Approach For Modi Script Character Recognition

Abstract

Character recognition is simple for humans, but developing software that is able to do so is a significant challenge. MODI is a language that has been neglected for a long time. We want to construct a character recognition system to read printed and handwritten MODI characters by scanning the papers, and our plan is to use this method. There is not much of a difference in the level of recognition accuracy achieved regardless of the quality of these articles. The established MODI Script Character Recognizer System (MSCR), which makes use of convolutional neural networks, presents challenges for the purpose of keeping a realistic character recognition system in place (CNN). The performance of this approach is evaluated by utilizing the dataset for Modi character recognition provided by IEEE Dataports. The CNN approach is utilized in order to give labels to each of the different Modi characters.