SAVITRIBAI PHULE PUNE UNIVERSITY, PUNE
A
PROJECT (STAGE-II) REPORT
ON
“Identification of Handwritten Devanagari Consonant Characters
using Deep Convolutional Neural Network “
SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE
AWARD OF THE DEGREE OF
MASTER OF ENGINEERING
IN
VLSI AND EMBEDDED SYSTEMS
By
Snehal S. Dholi
Seat No. 5089
UNDER THE GUIDANCE OF
PROF. Dr. S.T. Gandhe
SANDIP FOUNDATION'S
SANDIP INSTITUTE OF TECHNOLOGY & RESEARCH CENTRE, NASHIK
DEPARTMENT OF ELECTRONICS & TELECOMMUNICATION ENGINEERING
MAHIRAVANI, TRIMBAK ROAD NASHIK-422213
2021-2022
SANDIP FOUNDATION'S
SANDIP INSTITUTE OF TECHNOLOGY & RESEARCH CENTRE, NASHIK
Department of Electronics and Telecommunication Engineering
Mahiravani, Trimbak Road Nashik-422213, Maharashtra.
This is to certify that, this Project Stage-II report entitled “Identification of Handwritten
Devanagari Consonant Characters using Deep Convolutional Neural Network”
Submitted by Mrs Snehal S. Dholi for partial fulfillment of the requirement for the award
of the Master of Engineering in VLSI AND EMBEDDED SYSTEM as laid down by the
SAVITRIBAI PHULE PUNE UNIVERSITY, Pune. This is a record of her own work
carried out by her under my supervision and guidance during academic year 2021- 22
Place: - Nashik
Date: - 31/07/21
Exam No.: -5089
Prof. (Dr.) S. T. Gandhe Prof. Dr. Omkar Vaidya Prof. (Dr.) Omkar Vaidya
Guide Co-Guide ME Coordinator
E&TC Department E&TC Department E&TC Department
Prof. (Dr.) Gayatri M. Phade Prof. (Dr.) S. T. Gandhe
H.O.D Principal
E&TC Department SITRC, Nashik
SANDIP FOUNDATION'S
SANDIP INSTITUTE OF TECHNOLOGY & RESEARCH CENTRE, NASHIK
Department of Electronics and Telecommunication Engineering
Mahiravani, Trimbak Road Nashik-422213, Maharashtra.
This is to certify that, this Project Stage-II work entitled “Identification of Handwritten
Devanagari Consonant Characters using Deep Convolutional Neural Network” has
been successfully completed during the academic year 2021-22 by Mrs Snehal S.
Dholi .This work completed conforms to the standards laid down by Savitribai Phule Pune
University, Pune and has been completed in satisfactory manner as a fulfillment for the
Master of Engineering in VLSI AND EMBEDDED SYSTEM of Savitribai Phule Pune
University, Pune.
Place: - Nashik
Date: - 31/7/21
Exam No.: -5089
………………….. …………………..
EXAMINER – 1 EXAMINER - 2
ACKNOWLEDGEMENT
The work procedure in this report would not have been completed without the
encouragement and support of many people who gave their precious time and
encouragement throughout this period. First and foremost I would like to express my
deepest gratitude to my project guide Prof. Dr. S.T. Gandhe and Co-Guide Prof. Dr. Omkar
Vaidya for his invaluable support, guidance, motivation and encouragement throughout the
period this work was carried out. His readiness for consultation at all times, his concern and
assistance even with partial things has been extremely helpful. It was a great pleasure to
work under his guidance.
I am very much thankful to Prof. Dr. Omkar Vaidya (M.E. Coordinator) for giving
me many useful inputs and guidance for improving work throughout the entire period. His
suggestions with practical things have been extremely helpful. We are also grateful to Prof.
(Dr.) Gayatri M. Phade (Head of Department-E&TC Engineering) for her continuous
motivation, support in all aspects and comments to improve dissertation.
I am most grateful to our honorable Principal Prof.(Dr.) S. T. Gandhe for giving me
important comments to improve my dissertation sincerely thank to the entire team of staff
members, our college, our family, and those who knowingly and unknowingly have
contributed in their own way in completion of this dissertation.
Mrs. Snehal S. Dholi
Roll No : ETC/2015/VES/05
Seat No : 5089
Abstract
The Devanagari is an Indian originated script which is used in many official Indian
languages. There are 33 consonants, 13 vowels and 14 modifiers without any upper
and lower case characters in Devanagari script. More than 600 million users across
globe use Devanagari script in their daily routine. In this paper, we used 33
Devanagari Consonant having 200 images of each character as datasets for training.
We implemented proposed model by using Deep Convolutional Neural Network and
computed validation accuracy, validation loss and obtained confusion matrix.
We compared the proposed model with base model in terms of accuracy, precision,
recall and F1 score. The recognition accuracy for Devanagari consonant character is
recorded as 84.14% with our proposed model.
