INDUSTRIAL TRAINING REPORT

ON
"DEEP LEARNING"

Submitted in Partial Fulfillment of The R

Requirements
For The Award of The D Degree of

BACHELOR OF TECHNOLOGY
IN
COMPUTER SCIENCE ENGINEERING

Submitted By
NITIN SEN
Enrollment N
No. (0901CS203D07)

Department of Computer Science Engineering

Madhav Institute of Technology

&
Science,
Gwalior(M.P.)

JUN-JUL 2022
 CERTIFICATE

Certified that training work entitled "Industrial Training On Deep Learning" is a bonafied work
carried out in the seventh semester by Nitin Sen " In partial fulfillment for the award of the
degree of Bachelor of Technology in Computer Science Engineering from Madhav Institute of
Technology & Science during the academic year 2022-23.

Department Internship Coordinator

Prof. Hemlata Arya
Assistant Professor,CSE

Department Internship Coordinator

Prof. Jigyasa Mishra
Assistant Professor, CSE

Dr. Manish Dixit

H.O.D, CSE Deptt.
 DECLARATION
I hereby declare that I have completed my six weeks summer internship training at MITS Gwalior from
June 2022 to July 2022 under the guidance of Prof. Puneet Kumar Johari. I have declared that I have
learnt with full dedication during these six weeks of training and my learning outcomes fufill the
requirements of training for the award of degree of Bachelor of Technology (B.Tech) in CSE MITS
Gwalior.

Date : 22.11.22 Nitin Sen

0901CS203D07
 ACKNOWLEDGEMENT

I would like to acknowledge the contributions of the following people without whose help and
guidance this report would not have been completed.

I acknowledge the counsel and support of our training coordinator, Mrs. Jigyasa Mishra,
Assistant Professor, CSE Department, with respect and gratitude, whose expertise, guidance,
support, encouragement, and enthusiasm has made this report possible. Their feedback vastly
improved the quality of this report and provided an enthralling experience. I am indeed proud
and fortunate to be supported by him/her.

I am also thankful to Dr. Manish Dixit, H.O.D of Computer Science Engineering

Department, Madhav Institute of Technology & Science, Gwalior (M.P.) for his constant
encouragement, valuable suggestions and moral support and blessings.

Although it is not possible to name individually, I shall ever remain indebted to the faculty
members of Madhav Institute of Technology & Science, Gwalior (M.P.) for their persistent
support and cooperation extended during this work.

This acknowledgement will remain incomplete if I fail to express our deep sense of obligation
to my parents and God for their consistent blessings and encouragement.

Nitin Sen

0901CS203D07

I
 APPLICATIONS OF DEEP LEARNING

Applications of Deep Learning include:

 Web Search Engine: One of the reasons why search engines like google, bing etc work so
well is because the system has learnt how to rank pages through a complex learning
algorithm.

 Photo tagging Applications: Be it facebook or any other photo tagging application, the
ability to tag friends makes it even more happening. It is all possible because of a face
recognition algorithm that runs behind the application.

 Spam Detector: Our mail agent like Gmail or Hotmail does a lot of hard work for us in
classifying the mails and moving the spam mails to spam folder. This is again achieved by a
spam classifier running in the back end of mail application.

 Database Mining for growth of automation: Typical applications include Web-click data for
better UX, Medical records for better automation in healthcare, biological data and many
more.

 Applications that cannot be programmed: There are some tasks that cannot be programmed
as the computers we use are not modelled that way. Examples include Autonomous Driving,
Recognition tasks from unordered data (Face Recognition/ Handwriting Recognition),
Natural language Processing, computer Vision etc.

 Understanding Human Learning: This is the closest we have understood and mimicked the
human brain. It is the start of a new revolution, The real AI. Now, after a brief insight lets
come to a more formal definition of Machine Learning.

II
 PREFACE
The content of this report gives a detailed description of the I.T period activities. The I.T exercise,
logbook coupled with this report is primarily meant to satisfy a partial fulfillment of the requirements
for the award of a Bachelor's degree in Science of Electrical Engineering. The I.T exercise also enables
me to apply the concepts learnt at the university practically within the real life working environment.
The report has been divided as follows;

Chapter 1: This consists of a brief background about my industrial training project, the project
objectives and an overview about the internship training scope.

Chapter 2: This consists of the Literature review about computer vision and Deep learning, which are
very important in developing our product. In order to gain the best understanding of the project our
field advisor advised us to take a specialization course on Coursera offered by Prof Andrew Ng, thus
this chapter covers a review of material we learnt from that specialization.

Chapter 3: This consists of the methodology used in developing the project i.e. the chapter covers the
applications of knowledge acquired in the specialization course (chapter 2) in the development of the
Verec software system, it also covers all the tools and resources used in designing and training the
object detector.

Chapter 4: Contains the achievements, challenges, recommendations, and conclusion.

III
 TABLE OF CONTENTS
Abbreviations ………………………………………………………………………………………………….. v

Table of figures ………………………………………………………………………………………………...vi

Table of equations ……………………………………………………………………………………………...vi

CHAPTER 1: PROJECT REVIEW

1.1 Project Background …………………………………………………………………………………………. 1

1.2 Project Objectives ……………………………………………………………………………………………2

CHAPTER 2: LITERATURE REVIEW

2.1 Neural Networks and Deep Learning ………………………………………………………………………..3

2.2 Improving Deep Neural Networks: Hyper-parameter tuning, Regularization and Optimization ……………7

2.3 Structuring Machine learning Projects ……………………………………………………………………...11

2.5 Convolutional Neural Networks ……………………………………………………………………………13

CHAPTER 3: METHODOLOGY

3. 1 Introduction ………………………………………………………………………………………………..16

3.2 Design and Implementation of Verec object detector ……………………………………………………...16

3.3 Dataset and Features ………………………………………………………………………………………..17

3.4 Method: Open-Source Implementation of Yolo ……………………………………………………………18

3.5 Results & Discussion ……………………………………………………………………………………….19

3.6 Error analysis ……………………………………………………………………………………………….20

CHAPTER 4: CONCLUSION AND REMARKS.

4.1 Achievements ……………………………………………………………………………………………....21

4.2 Conclusion ………………………………………………………………………………………………....21

IV
 ABBREVIATIONS

Adam Adaptive Moment Estimation.

NN Neural Network
ANN Artificial Neural Network
CNN Convolutional Neural Network
FPS Frames per second
ResNet Residual Networks
AP Average Precision
MAP Mean Average precision
Yolo You only look once
LDC Least Developed Countries
RTA's Road traffic Accidents
WHO World health Organization
AI Artificial Intelligence
API Application programming interface
Max Maximum
UN United Nations
ICT Information Communication Technology
SDG Sustainable Development goals
RNN Recurrent Neural Networks
LSTM Long Short Term Memory networks
IoU Interception over Union
Conf Configuration
Thres Threshold
GPU Graphic Processor Unit
Relu Rectified Linear Unit
RMSprop Root Mean Square Propagation
Norm Normalization
ML Machine Learning
Dev Development
3D Three-dimensional
Convs Convolutions
V Version
SIMD Single Instruction Multiple Data
SAAS Software as a service
CIFAR Canadian Institute for Advanced Research

V
 TABLE OF FIGURES

FIGURE 1, NEURAL NETWORK [6] ………………………………………………………………… 3

FIGURE 2: BINARY CLASSIFICATION [24] ……………………………………………………….. 4
FIGURE 3: GLOBAL OPTIMUM [25] ………………………………………………………………… 6
FIGURE 4: BIAS AND VARIANCE [9] ……………………………………………………………….. 7
FIGURE 5: BATCH AND MINI-BATCH GRADIENT DESCENT [6] ………………………………...9
FIGURE 6: ERROR ANALYSIS [6] …………………………………………………………………... 12
FIGURE 7: CNN OVERVIEW [14] …………………………………………………………………… 13
FIGURE 8: OBJECT DETECTION [6] ………………………………………………………………... 14
FIGURE 9: YOLO OBJECT DETECTOR PIPELINE [15] …………………………………………… 15
FIGURE 10: GRID CELLS IN OBJECT DETECTION [27] ………………………………………….. 17
FIGURE 11: OPEN IMAGE DATASET LABELS [29] ………………………………………………. 17
FIGURE 12: DATASET ANNOTATION FORMAT [20] …………………………………………….. 18
FIGURE 13: ANNOTATION ……………………………………………………………………………18
FIGURE 14: IMAGE …………………………………………………………………………………….18
FIGURE 15: HELMET DETECTION (IMAGE FROM SAFEBODA) ………………………………...19
FIGURE 16:MAP ………………………………………………………………………………………..20

Table of equations
EQUATION 1: NEURAL NETWORK [6] ……………………………………………………………... 3
EQUATION 2, LOGISTIC REGRESSION …………………………………………………………….. 5
EQUATION 3, SIGMOID FUNCTION ………………………………………………………………… 5
EQUATION 4: MEAN SQUARE ERROR ……………………………………………………………... 5
EQUATION 5: LOGISTIC REGRESSION LOSS FUNCTION. ………………………………………. 5
EQUATION 6: COST FUNCTION …………………………………………………………………….. 5
EQUATION 7: UPDATING WEIGHTS W ……………………………………………………………. 6
EQUATION 8: UPDATING WEIGHTS B …………………………………………………………….. 6
EQUATION 9: L1 REGULARIZATION MATRIX …………………………………………………… 6
EQUATION 10: LI REGULARIZATION ……………………………………………………………… 8
EQUATION 11: L2 REGULARIZATION MATRIX ………………………………………………….. 8
EQUATION 12: L2 REGULARIZATION …………………………………………………………….. 8
EQUATION 13: F1 SCORE …………………………………………………………………………… 11
EQUATION 14: OUTPUT OF YOLO [18] …………………………………………………………... 17

VI