Recognition of Teeline Shorthand using Deep Learning
Shivaprakash1, Vishwanath C Burkpalli2, B. S. Anami3

1Mr. Shivaprakash, Department of Computer Science and Engineering Government Engineering College, Devagiri, Haveri, India.
2Dr. Vishwanath C Burkpalli, Department of Information Science and Engineering P.D.A. College of Engineering Gulbarga, India.
3Dr. B. S. Anami, Principal, K.L.E. Institute of Technology Hubli, India.
Manuscript received on December 16, 2019. | Revised Manuscript received on December 20, 2019. | Manuscript published on January 10, 2020. | PP: 1242-1247 | Volume-9 Issue-3, January 2020. | Retrieval Number: C8693019320/2020©BEIESP | DOI: 10.35940/ijitee.C8693.019320
Open Access | Ethics and Policies | Cite | Mendeley
© 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: Recognition of the Teeline shorthand is the interesting research problem. The accurate recognition of the Teeline patterns which helps the journalists to summarize the VIPs talk during the press meets which reduces the writing time. In order to address the above issues, the proposed method addresses the automatic recognition of the Teeline symbols which is used to provide alphabets as output. The framework consists of obtaining the input image, pre-processing it, perform the background subtraction on grayscale image and prepare the dataset for recognition. To prepare the dataset, the letters are written using the digital pen, it generates the pixelsand it is recorded in the Android application. Later these pixels are used to convert them into the image dataset. The prepared database is fed to the manually designed eleven layered Convolutional Neural Network (CNN). CNN is designed in such a way that it accepts the input Teeline alphabet and it able to say the equivalent English alphabet based on the training process. It has been carried out on the MATLAB 2018b and achieved the accuracy of 95%. 
Keywords: Teeline shorthand, CNN, Matlab, Accuracy.
Scope of the Article: Deep Learning