HANDWRITTEN TELUGU COMPOUND CHARACTER PREDICTION USING CONVOLUTIONAL NEURAL NETWORK

HANDWRITTEN TELUGU COMPOUND CHARACTER PREDICTION USING CONVOLUTIONAL NEURAL NETWORK

Dr. C. Sunitha Ram, U.T.S. Abhiroopika, S. Chinmayee

SCSVMV UNIVERSITY

Abstract: Whether in Telugu or another language, handwriting recognition has been a difficult job for a very long time. Handwritten characters frequently have many curves, particularly in Indian languages. With the development of machine learning, handwritten recognition has long been an open task. However, there are still many difficulties because feature extraction is a challenging job because characters are more prevalent in Indian languages. Pick a handwritten Telugu compound character named Guninthalu (a character made up of Telugu vowels and consonants) to be recognised in this essay. Since each character in Guninthalu is nearly identical to the others, Telugu. It's difficult to classify things. Although there are many machine learning methods, achieving accuracy is the main obstacle. Consequently, applying deep learning. Convolutional neural networks are being used to build a machine-learning model for Telugu Handwriting Gunithalu. Create a dataset, and the IEEE Data port will have it accessible.

Keywords: Telugu Character Prediction, Convolutional Neural Network, Deep Learning, Handwritten Letter Prediction.