Preserving the Legacy of Ancient Tamil Script with Deep Learning and Fuzzy C Means Algorithm: Intelligent Approach to Digiitization

Handwritten character recognition has become a focus of research since the emergence of deep learning technologies. This research objective is to develop and establish an intelligent model to automate an existing manual system for digitising Ancient Tamil script in order to save time and to preserve antique data. In contrast to Western languages, Tamil script is an old popular script in India that lacks institutional digitising methods. In this paper, To segment and extract the features for further classification, a fuzzy C-means algorithm is employed. This research uses the most appropriate strategies to improve recognition rates and configures a convolutional neural network for effective ancient inscription text recognition. This method uses the dataset of estampage images of Ancient Copper plate inscriptions that contains 65 classes of Ancient Tamil characters and each class has 50 different images. In terms of accuracy and training duration, this method has shown to be more effective for Ancient Character recognition.