CNN Based Character Recognition and Classification in Tamil Palm Leaf Manuscripts

Palm leaf manuscripts are extremely important as they have a rich source of information. As a result, simple access to ancient manuscripts must be provided to share this information with the rest of the world and to promote future study into ancient literature. This study deals with Convolutional Neural Network (CNN)-based Optical Character Recognition (OCR) system for accurately digitizing and identifying the characters for Tamil palm leaf manuscripts. The convolution layer, pooling layer, activation layer, fully connected layer, and classifier of the convolutional neural network is employed in this article. Palm-leaf manuscripts were scanned and the scanned images are used to generate the character set database. The database is divided into 67 separate classes, each of which contains roughly 100 individual samples. OCR recognition of the palm leaf manuscripts and problems associated with this are illustrated. A working example of the character recognition method for Tamil palm-leaf manuscript was implemented using the CNN model. The CNN model was found to have a better recognition rate. The prediction rate and accuracy are great because of the large number of features retrieved for each layer of CNN.