Manuscript Title:

OFFLINE WRITER IDENTIFICATION USING CNN AND RNN: A DEEP LEARNING APPROACH

Author:

DEEPA BENDIGERI, Dr. GOPAL A. BIDKAR, Dr. JAGADEESH D.PUJARI

DOI Number:

DOI:10.17605/OSF.IO/KADQV

Published : 2022-12-10

About the author(s)

1. DEEPA BENDIGERI - Assistant Professor, Department of Information Science and Engineering, SDM College of Engineering and Technology, Dharwad, India.
2. Dr. GOPAL A. BIDKAR - Professor Department of Electronics and Communication Engineering, SDM College of Engineering and Technology, Dharwad, India.
3. Dr. JAGADEESH D.PUJARI - Professor and Head, Department of Information Science and Engineering, SDM College of Engineering and Technology, Dharwad, India.

Full Text : PDF

Abstract

Pattern recognition and artificial intelligence have a lot to say about how to identify a writer. To identify the writers, conventional methods rely on hand-craft features. Convolutional Neural Networks (CNNs) have recently emerged as the most effective method for classifying Writers. The Recurrent Neural Network (RNN) models the spatial relationship between the sequences of fragments in order to make the local fragment features better at telling them apart. So, this work, suggests CNN and RNN models for identifying offline writers. The proposed method used IAM dataset of offline handwritten english script for experimentation. The proposed method achieved 90% accuracy to classify 690 writers using CNN and 93% of accuracy using RNN. The proposed method can generate efficient and robust writer identification based on different scale, different orientation and both.


Keywords

Writer Identification, Offline Analysis, Convolution Neural Networks, Deep Learning Recurrent Neural Network, IAM Database.