Real time Air Writing and Recognition of Tamil Characters with voice integration using deep learning

In this paper, a real time air writing and recognition system for Tamil alphabets using deep learning has been proposed. Air-writing is writing characters in the free space using fingers. The trajectory of the air written characters is mapped by using Media Pipe Function. The obtained trajectory is then pre-processed and given to Dense Net 121 which is a type of Convolutional Neural Network (CNN) model widely used for pattern matching along with the dataset from HP labs which contains 18215 images for 11 uyir ezhuthugal, 18 mei ezhuthugal,5 vada mozhi ezhuthugal and 1 aayudha ezhuthu. The model which is trained and obtained a maximum training and validation accuracy of 83.58% and 93.53% respectively with minimum training and validation loss of 478.45% and 19.7% respectively with F1 score of 0.9355.