Enhancing Gujarati Numeral Recognition With Convolutional Neural Networks

In the current scenario, digitization of handwritten documents is crucial. The process of digitizing handwritten documents has become extremely popular in recent decades. In a country like India, there are 22 official languages and 13 different script languages. Therefore, it was very difficult to digitize this handwriting for further editing. This article attempts to classify handwritten Gujarati digits to improve accuracy in further processing. There are 10 different categories of Gujarati numbers and the classification process is 94% accurate. This article introduces a method for offline recognition of handwritten Gujarati digits. The best result is achieved by utilizing convolutional neural networks on the developed Gujarati dataset.