Handwritten Text Recognition: With Deep Learning and Android
Shubham Sanjay Mor1, Shivam Solanki2, Saransh Gupta3, Sayam Dhingra4, Monika Jain5, Rahul Saxena6
1Shubham Sanjay Mor, Department of Computer Science and Engineering, Manipal University, Jaipur (Rajasthan), India.
2Shivam Solanki, Department of Computer Science and Engineering, Manipal University Jaipur (Rajasthan), India.
3Saransh Gupta, Department of Computer Science and Engineering, Manipal University, Jaipur (Rajasthan), India.
4Sayam Dhingra, Department of Computer Science and Engineering, Manipal University Jaipur (Rajasthan), India.
5Monika Jain, Department of Information Technology, Manipal University, Jaipur (Rajasthan), India.
6Rahul Saxena, Department of Information Technology, Manipal University, Jaipur (Rajasthan), India.
Manuscript received on 22 April 2019 | Revised Manuscript received on 01 May 2019 | Manuscript Published on 05 May 2019 | PP: 172-178 | Volume-8 Issue-2S2, May 2019 | Retrieval Number: B10370182S219/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: This research paper offers a new solution to traditional handwriting recognition techniques using concepts of Deep learning and computer vision. An extension of MNIST digits dataset called the Emnist dataset has been used. It contains 62 classes with 0-9 digits and A-Z characters in both uppercase and lowercase. An application for Android, to detect handwritten text and convert it into digital form using Convolutional Neural Networks, abbreviated as CNN, for text classification and detection, has been created. Prior to that we pre-processed the dataset and applied various filters over it. We designed an android application using Android Studio and linked our handwriting text recognition program using tensorflow libraries. The layout of the application has been kept simple for demonstration purpose. It uses a protobuf file and tensorflow interface to use the trained keras graph to predict alphanumeric characters drawn using a finger.
Keywords: E.G – For Example, NN – Neural Network, RNN – Recurrent Neural Network, CNN – Convolutional Neural Network, EMNIST – Extended NIST Dataset.
Scope of the Article: Deep Learning