Handwritten Character Recognition of Kannada Language Using Convolutional Neural Networks

Abstract: Handwritten Character Recognition of Kannada Language using Convolutional Neural Network is the project aimed at preserving the handwritten script in digital format, particularly for Kannada language, since Kannada is a language that is spoken by almost all of the residents of Karnataka. Kannada language has 49 base characters which include 15 vowels and 34 consonants. The project focuses on converting handwritten characters or sentences by recognizing them and converting them into digital format using Convolutional Neural Network (CNN) technology. To determine the model’s effectiveness, it must first be trained, then validated and ultimately tested. The model gave a prediction performance of 97 % for the testing set. This result highlights the potential of this particular project to significant improvements in the accessibility to historical records and the streamlining of administrative processes through the successful implantation of the handwritten character recognition system.

Keywords: Convolutional Neural Network, Handwritten Character Recognition, Artificial Intelligence, Optical Character Recognition.