310255: Seminar and Technical

Communication
(Guidelines and Log Book)

Third Year Computer Engineering

Year 2022 - 2023

Seminar ID: 19

Name of Student: SHASHANK BIRAJDAR

Mobile No.: 895663457 e- Mail ID: shashank.birajdar_comp2020@pccoer.in

Seminar Title : Object detection for vehicles and computer components

Seminar Guide : Mr. Mahendra Salunke

Area of the Seminar: Image processing

Department of Computer Engineering

Pimpri Chinchwad Educational Trust’s
Pimpri Chinchwad College Of Engginering & Research, Ravet

Savitribai Phule Pune University

(2022-2023)
 General Instructions

1. Students should enter the correct information in the work book.

2. Get all entries verified by respective seminar guide. No changes are to be made

without seminar guide’s permission.

3. Students should report to their respective guides as per the schedule and the

visit log is to be maintained in the work book.

4. Follow all deadlines and submit all documents strictly as per prescribed

formats.

5. The work book should be produced at the time of all discussions and

presentations.

6. The work book must be submitted to Seminar coordinator/ guide/ department

/ College after successful examination.

7. All documents and reports are to be prepared in Latex only (All the formats
specifications provided adheres to MS Word but consequently applicable to final
seminar report published using Latex)

8. Submit hard as well as soft copy as per guidelines

i
This booklet is supportive document to rules and a regulation provided by affiliated
university curriculum providing recommendations, guidelines and is record of all
related activities associated with seminar. This booklet is provided with the genuine
intent to bring uniformity and to systematize the seminar work and to keep the
audit of the work undergone by each student.

Work Book Development Project

Project Institution Department of Computer Engineering
Matoshri College of Engineering and Research Centre, Nashik

Support & Guidance Dr. Gajanan K. Kharate,

Principal, Matoshri College of Engineering and Research Centre,
Nashik
Concept and Design Dr. Varsha. H. Patil
BoS Coordinator Computer Engineering , SPPU, Pune
Vice Principal, Matoshri College of Engineering and Research Centre,
Nashik
Coordinator Mrs. Swati A. Bhavsar
Assistant Professor,
Matoshri College of Engineering and Research Centre, Nashik

Technical 1. Dr. Madhavi Pradhan

Committee Members
2. Dr. Parikshit Mahalle
3. Mr. Niranjan L. Bhale
4. Dr. Neeta Deshpande
5. Mr. Ranjit Gawande
6. Ms. Sharmila Wagh

Date 21st April 2017

Version No. 2.0

Copyright Diary No.- 3869/2017-CO/L

(All rights reserved )

(For circulation at BoS Computer Engineering, Savitribai Phule Pune University only)

ii
 Savitribai Phule Pune University, Pune
Computer Engineering
Program Educational Objectives
1. To prepare globally competent graduates having strong fundamentals, domain knowledge, updated
with modern technology to provide the effective solutions for engineering problems.
2. To prepare the graduates to work as a committed professional with strong professional ethics and
values, sense of responsibilities, understanding of legal, safety, health, societal, cultural and
environmental issues.
3. To prepare committed and motivated graduates with research attitude, lifelong learning, investigative
approach, and multidisciplinary thinking.
4. To prepare the graduates with strong managerial and communication skills to work effectively as
individual as well as in teams.
Program Outcomes
Students are expected to know and be able –
1. To apply knowledge of mathematics, science, engineering fundamentals, problem solving skills,
algorithmic analysis and mathematical modeling to the solution of complex engineering problems.
2. To analyze the problem by finding its domain and applying domain specific skills
3. To understand the design issues of the product/software and develop effective solutions with
appropriate consideration for public health and safety, cultural, societal, and environmental
considerations.
4. To find solutions of complex problems by conducting investigations applying suitable techniques.
5. To adapt the usage of modern tools and recent software.
6. To contribute towards the society by understanding the impact of Engineering on global aspect.
7. To understand environment issues and design a sustainable system.
8. To understand and follow professional ethics.
9. To function effectively as an individual and as member or leader in diverse teams and interdisciplinary
settings.
10. To demonstrate effective communication at various levels.
11. To apply the knowledge of Computer Engineering for development of projects, and its finance and
management.
12. To keep in touch with current technologies and inculcate the practice of lifelong learning.

Program Specific Outcomes (PSO)

A graduate of the Computer Engineering Program will demonstrate-
PSO1: Professional Skills-The ability to understand, analyze and develop computer programs in the areas
related to algorithms, system software, multimedia, web design, big data analytics, and networking for
efficient design of computer-based systems of varying.
PSO2: Problem-Solving Skills- The ability to apply standard practices and strategies in software project
development using open-ended programming environments to deliver a quality product for business
success.

PSO3: Successful Career and Entrepreneurship- The ability to employ modern computer languages,
environments, and platforms in creating innovative career paths to be an entrepreneur, and a zest for higher
studies.

iii
 Prologue

Seminar is the first formal curricular activity at the UG level, where students are supposed
to exhibit their skills and knowledge by undertaking the study of the chosen topics. For
standardization in the process of Seminar conduction, an effort to provide comprehensive and
meticulous guidelines helping the learners to perform with respect to certain processes and
evaluation criteria.
The logbook will surely help the learner from the very first step of topic selection to the
final seminar delivery. Proper recording of the activities necessarily maintains the track of
progress of the learner along with neat and clear planning helping to proceed on the right path
and proper documents preparation. As per the individual learner’s domain interest the selected
topic can be explored with determined perspective and definite methodology helping the learner
to develop scientific and methodical approach in the study. In the course of the topic exploration
various skills are built, directly and indirectly contributing to the development of learner.
The documentation provided in the form of the logbook will help to standardize the process
with phenomenal transparency in evaluation guidelines, giving fair idea to learner and evaluator,
minimizing the possibility to err. The documented logbook will hopefully answer even the slightest
queries those may arise during the whole process of the activity conduction during the semester.
So, it is our joint responsibility to stick to the basics to help the learner in character building not
solely aiming at the grade in performance but aiming at all-round development in this regard.

Dr. Varsha H. Patil

Coordinator, Board of Studies (Computer Engineering)
SPPU, Pune 22nd February 2017.

iv
 Table of Contents

Sr. Description Page

No. No.

General Instructions i

Work Book Development Project ii

Computer Engineering PEO’s, PO’s and PSO iii

Prologue iv

1. About Seminar 1
a. Objectives and Outcomes
b. Guidelines for Selection of Seminar Topic
c. Guidelines for Evaluation
2. Copy of Synopsis 3

3. Review and Visit Log 4

4. Seminar Evaluation Sheet 5

5. Paper Presentation / Participation at Conference 6

6. Rubrics 7

Annexures

i. Format for Synopsis 8

ii. Format for Seminar Report 9

iii. Format for Cover Page 11

iv. Format for Certificate 12

v. Format for Report Documentation 13

1. About Seminar

The word seminar is derived from the Latin word seminarium, meaning "seed
plot". It refers to a course of intense study relating to the student's major
intended for the improvement of technical knowledge of student. The ability
to articulate ideas is an important life skill which will be required outside the
academic world in the world of work, for interviews, consulting experts,
getting and understanding advice and giving work presentations etc. Seminars
give practice in these general skills and help students to develop confidence.
It is an important way of learning - by discussing and questioning issues,
students can clarify their own ideas and also learn from each other. (Ref:
https://en.wikipedia.org/wiki/Seminar)
Keeping this in mind each student of Third Year Computer Engineering
has to deliver the seminar under the head “SEMINAR AND TECHNICAL
COMMUNICATION” that is Term Work of 50 marks in second semester.
As per the individual learner’s domain interest the selected topic can
be explored with determined perspective and definite methodology helping
the learner to develop scientific and methodical approach in the study. In the
course of the topic exploration various skills are built, directly and indirectly
contributing to the development of learner.
To aid both student and faculty this booklet provides the guidelines for
preparation of topic, report, presentation, evaluation.

a. Objectives and Outcomes

Objectives -

• To explore the basic principles of communication (verbal and non-

verbal) and active, empathetic listening, speaking and writing
techniques.
• To expose the student to new technologies, researches, products,
algorithms, services

Outcomes -
On completion of the course, student will–
• be able to be familiar with basic technical writing concepts and terms,
such as audience analysis, jargon, format, visuals, and presentation.
• be able to improve skills to read, understand, and interpret material
on technology.
• improve communication and writing skills
b. Guidelines for selection of Seminar Topic
• Each student will select a topic in the area of Computer Engineering
and Technology preferably keeping track with recent technological
trends and development beyond scope of syllabus avoiding repetition
in consecutive years.
• The topic must be selected in consultation with the institute guide.

Seminar Log Book Third Year Computer Engineering, SPPU, Pune Page 1
 • Each student will make a seminar presentation using audio/visual aids
for duration of 20-25 minutes and submit the seminar report prepared
in Latex only.
• Active participation at classmate seminars is essential.
• Softcopy (CD) must include copy of synopsis, report, PPT, reference
material and related.

c. Recommended Guidelines for Evaluation

Panel of staff members along with a guide would be assessing the seminar
work based on these parameters-
• Topic
• Contents and Presentation
• regularity, Punctuality and Timely Completion
• Question and Answers
• Report, Paper Presentation/Publication
• Attendance and Active Participation.

(Kindly note that these guidelines provided for selection, evaluation,

presentation and documentation are recommended to follow. However it is
suggested to refer the guidelines prescribed in respective course of syllabus
by SPPU)

Seminar Log Book Third Year Computer Engineering, SPPU, Pune Page 2
 2. Copy of Synopsis (Annexure i)

Synopsis

Name of the Student: Shashank Birajdar

Roll No: TYCOB03 Branch: CSE

Mobile no:8956634575

Email ID: Shashank.birajdar_comp2020@pccoer.in

Seminar Guide: Mr. Mahendra Salunke

Signature of Student Signature of seminar guide

1. Title of the topic: Object detection for vehicles and computer components

2. Area of topic: COMPUTER VISION

3. Abstract:
With the development of artificial intelligence, computer vision technology that
simulates human vision has received widespread attention. Based on the current commonly
used method of computer vision technology-deep learning, this paper outlines the development
of deep learning models, and determines the inflection point of the development of the
introduction of convolutional neural networks. Not only that, convolutional neural networks
combined with deep learning have developed rapidly and are widely used in the field of
computer vision. Therefore, the structure and development process of the convolutional neural
network are further analyzed. Subsequently, the three major difficulties in the field of computer
vision-image classification, object detection, and image segmentation are compared and
analyzed, and their differences and connection analysis are drawn. Finally, an overview of the
difficulties to be optimized in computer vision, and computer vision technology It will continue
to develop as a research hotspot.

Computer vision as a discipline is at the cutting edge of science. As with any frontier,
it is thrilling and chaotic, with often no trustworthy authority to turn to. Numerous beneficial
concepts lack a theoretical foundation, and some theories are rendered ineffective in reality;
developed regions are widely dispersed, and often one seems totally unreachable from the
other.

Recently, computer vision has gained traction and popularity as a consequence of the
many applications it has found in areas like health and medical, sports and entertainment,
automaton design, and self-driving cars. Many of these applications rely on visual recognition
tasks such as image order, restriction, and identification. Recent advances in Convolutional
Neural Networks (CNNs) have resulted in an extraordinary performance in these best-in-class
visual recognition assignments and frameworks, demonstrating the power of Convolutional
Neural Networks.
Briefs about Contents:
CNN, or convolutional neural network, is a kind of neural network
model that is designed for dealing with two-dimensional image data, although it may also be
used to deal with one dimensional and three-dimensional data. Convolution is accomplished
via the use of a channel (a small matrix whose size may be chosen). In this channel, which
travels the whole picture network, the task is to reproduce the image's features by utilizing the
pixel values that were first used.
The learning algorithms helps in creating object and image detection
systems based on the instances and experiences. The machine learning algorithms renders
enormous capabilities for integration and synthesis of vision algorithms and models. Object
detection and tracking is still open challenge in computer vision, even though alignment with
machine learning algorithms and open source libraries have tendered exciting results.
Moreover, the machine learning output quality depends on the predictive accuracy, recall, and
precision. The data collected from the environment dictates the strategies that a learning system
would use so to improve the performance.
The basic concepts required to understand what Mask R-CNN is and
how it works: Convolutional Neural Networks (CNN) Region-Based Convolutional Neural
Networks (R-CNN) Faster R-CNN with Region Proposal Networks (RPN) Mask R-CNN and
how it works Example projects and applications Mask R-CNN Demo Sample To understand
the differences between Mask RCNN, Faster RCNN vs. RCNN, we first have to understand
what a CNN is and how it works. What is a Convolutional Neural Network (CNN)? A
Convolutional Neural Network (CNN) is a type of artificial neural network used in image
recognition and processing that is optimized to process pixel data. Therefore, Convolutional
Neural Networks are the fundamental and basic building blocks for the computer vision task
of image segmentation (CNN segmentation). The Convolutional Neural Network Architecture
consists of three main layers: Convolutional layer : This layer helps to abstract the input image
as a feature map via the use of filters and kernels. Pooling layer : This layer helps to down
sample feature maps by summarizing the presence of features in patches of the feature map.
Fully connected layer:  Fully connected layers connect every neuron in one layer to every
neuron in another layer.
3.1 Introduction

Convolutional neural networks help ML models see by fractionating images into

pixels. Each pixel is given a label or tag. These labels are then collectively used to carry out
convolutions, a mathematical process that combines two functions to produce a third function.
Through this process, convolutional neural networks can process visual inputs.

To see images just like a human would, neural networks execute convolutions and examine the
accuracy of the output in numerous iterations. Just like humans would discern an object far
away, a convolutional neural network begins by identifying rudimentary shapes and hard edges.
Once this is done, the model patches the gaps in its data and executes iterations of its output.
This goes on until the output accurately ‘predicts’ what is going to happen.

While a convolutional neural network understands single images, a recurrent neural network
processes video inputs to enable computers to ‘learn’ how a series of pictures relate to each
other. In 2015, technology leader Google rolled out its instant translation service that leverages
computer vision through smartphone cameras. Neural Machine Translation, a key system that
drives instantaneous and accurate computer vision-based translation, was incorporated into
Google Translate web results in 2016. Not to be left behind, technology giant Meta (earlier
known as Facebook) is also dabbling in computer vision for various exciting applications. One
such use is the conversion of 2D pictures into 3D models. YOLO, which stands for You Only
Look Once, is a pre-trained object detection model that leverages transfer learning. You can
use it for numerous applications, including enforcing social distancing guidelines.

As a computer vision solution, the YOLO algorithm can detect and recognize objects in a visual
input in real-time. This is achieved using convolutional neural networks that can predict
different bounding boxes and class probabilities simultaneously.
SentioScope is a fitness and sports tracking system developed by Sentio. It primarily operates
as a player tracking solution for soccer, processing real-time visual inputs from live games.
Recorded data is uploaded to cloud-based analytical platforms.
Literature review:
Research article Objective/Proposed Methods/Techniques Datasets Relevant
(Author/Year) work findings/Limitations
identified

Review of Intelligent To detect the objects RCNN + ATT Common Object detection in cars
Driving Pedestrian frames YOLOv3 pedestrian and surveillance camera:
Detection Based on And analyze process detection High velocity motions
Deep Learning according to it datasets can not be efficiently
(Di Tian , Yi Han) now detected
Published 21 July 2022 include
Caltech ,
KITTI ,
CityPersons
, TUD , and
EuroCity

Unified, Real-Time Road lane detection YOLO Detection System TUD, Real time object
Object Detection Caltech detection in sports such as
(Joseph Redmon∗ , crickets and golf for
Santosh Divvala) Tracking the object

Object Detection and intelligent driving and CNN MS COCO Motion sensing and
Tracking using Deep security monitoring SDS-RCNN dataset pedestrian detection:
Learning and Artificial quality of hardware can
Intelligence for Video matter
Surveillance
Applications
(Dulari Bhatt 1 , Chirag
Patel)

ARTIFICIAL Occupancy detection, YOLOv3 CIFAR-10 Poor quality of dataset

INTELLIGENCE IN traffic detection, Support Vector Machine IMDB- and input images is
COMPUTER VISION tracking, classification (SVM) Wiki ineffictive
Published Online May and counting. Bayesian Networks Dataset
2021 in IJEAST

Domain Adaptation for Pedestrian detection RCNN + ATT EuroCity

Pedestrian Detection using CNN and YOLOv3 MS COCO
J. Xu, D. Vazquez, identifying the object dataset
A.M. L ´ opez, J. Mar ´ travelling
´ın and D. Ponsa,
Learning a Multiview
Part-based Model in
Virtual World for
Pedestrian Detection.
In IEEE
 3.2 Objectives/Motivation

In Computer Vision (CV), ML performs a significant role to

extract crucial information from images. CV successfully contributes to
multiple domains, surveillancesystem, optical character recognition, robotics,
suspect detection, and many more. The direction of CV research is going
toward healthcare realm, medical imaging (MI) is the emerging technology,
play a vital role to enhance image quality and recognized critical features of
binary medical image, covert original image into grayscale and set the
threshold values for segmentation.

3.3 Methodology

Machine Learning Techniques for Image Processing Image

processing (IP)means process image through a digital computer. Utilization of
algorithms for enhanced qualityof an image, on the bases of this system, can
extract crucial information from the image [22, 23]. There are two main types
namely: analogous and digital image processing (DIP) [24, 25].In IP, get input
as an image and after process system retrieves output in the form of an image.
Mentioning the three important key factors of IP which are discussed below: •
Start with the image acquisition • Manipulating and analyzing the image •
Output in the form of an image which is based on analyzing.
Faster R-CNN advances this stream by learning the attention
mechanism with a Region Proposal Network and Fast R-CNN architecture.
The reason why “Fast R-CNN” is faster than R-CNN is because you don’t
have to feed 2’000 region proposals to the convolutional neural network every
time. Instead, the convolution operation is done only onceper image, and a
feature map is generated from it. Furthermore, Faster R-CNN is an optimized
form of R-CNN because it is built to enhance computation speed (run R-CNN
much faster).
You Only Look Once (YOLO) is one of the most popular model
architectures and object detection algorithms. It uses one of the best neural
network architectures to produce highaccuracy and overall processing speed,
which is the main reason for its popularity. If we search Google for object
detection algorithms, the first result will be related to the YOLO model.
YOLO algorithm aims to predict a class of an object and the bounding box that
defines the object location on the input image. It recognizes each bounding
box using four numbers:

• Center of the bounding box

• Width of the box
• Height of the box
References / Bibliography

• https://www.ibm.com/in-en/topics/computer-vision

• https://ijcrt.org/papers

• S. Zhang, J. Yang, and B. Schiele, “Occluded pedestrian detection through

guided attention in

• CNNs,” in Proceedings Of the Computer Vision And Pattern

Recognition, pp. 6995–7003, SaltLake City, UT, USA, June 2018.

• J. Xu, D. Vazquez, A.M. L ´ opez, J. Mar ´ ´ın and D. Ponsa, Learning a

Multiview Part-based Modelin Virtual World for Pedestrian Detection.
In IEEE

• S. Lin, Y. Zhang, C. Hsu, M. Skach, E. Haque, L. Tang, et al., "The

Architectural Implications of Autonomous Driving: Constraints and
Acceleration", Proceedings of the Twenty-Third International Conference on
Architectural Support for Programming Languages and Operating Systems, pp.
751-766, 2018.

Seminar Log Book Third Year Computer Engineering, SPPU, Pune Page 3
 3. Review and Visit Log

Signature of
Sr.
Date Details of Discussion/ Remark Guide/Seminar
No.
Incharge

25.7.2022
1 to Topic Finalization
5.8.2022
8.8.2022
Preparing draft of proposal of
2 to
seminar
12.8.2022
16.8.2022
3 to Submission of Proposal
20.8.2022
22.8.2022
4 23.8.2022 Review
24.8.2022

22.9.2022
5 Review
23.9.2022

25.8.2022
6 to Preparing for presentation
3.9.2022
5.9.2022
7 to Preparing for presentation
17.9.2022
19.9.2022
Seminar Report rough draft
8 to
Preparation
30.9.2022

9 30.9.2022 Seminar Report finalization

13.10.2022
10 Seminar Report submission
14.10.2022

Seminar Log Book Third Year Computer Engineering, SPPU, Pune Page 4
 4. Seminar Evaluation Sheet (Internal)
Table 1.1 Evaluation Sheet
Sr. Conten Paper
No. ts Attendanc Publication
Punctuality
and e and
and Timely Semin Question
Participatio
Presen and n
Completion ar and Total
tation Active at Conference
(following Report Answers
(Table participati (Bonus)
of deadline)
1.2) on

25 05 10 05 05 05 50

1.

# To be
filled by guide/ authorities
Whether the seminar is delivered as per
schedule(yes/ no):
(If no, mention the reason)

Table 1.2 Contents and Presentation

Slide Layout Verbal Skill Confidence Eye Contents

Contact Total

5 5 5 5 5 25

Name and Signature of Evaluation Committee:

1. Prof.

2. Prof.

Signature of
Guide: HOD: Mrs. Archana Chougule
Mr. Mahendra Salunke
(Refer Rubrics – page number 08)

\ Seminar Log Book Third Year Computer Engineering, SPPU, Pune Page 5
 5. Paper Publication/ Participation at Conference

Certificates/
Sr. Name of Organizer Date Prizes
No. won if any
1.

2.

3.

4.

Attach attested copy of certificate(s)

Seminar Log Book Third Year Computer Engineering, SPPU, Pune Page 6
 6. Rubrics
A) Contents and Presentation

Grade Excellent Very Good Fair Poor

(Grade Point) (10-9) (6-8) (3-5) (1-2)
Parameter

Slide Preparation

Verbal Skills

Confidence

Eye Contact

Contents

B) Overall performance

Grade Excellent Very Good Fair Poor

(Grade Point) (10-9) (6-8) (3-5) (1-2)
Parameter

Punctuality and Timely

Completion
Question and Answers

Attendance and Active

Participation

Seminar Report

Paper publication &

presentation

Seminar Log Book Third Year Computer Engineering, SPPU, Pune Page 7
Annexure i: Format for Synopsis

1) Cover Page:
Name of the Student:
Roll No: Branch: Email ID:
Mobile:

Title of the topic:

Area of topic
Abstract:
Abstract should be of approximately 200-300 words giving brief
introduction about the topic along with scope.
2) Briefs about Contents:
The contents shall follow abstract indicating the topics, sub
topics under consideration not exceeding two pages.
3) Applications areas, if any:
4) References / Bibliography
List of books/ web/ Journal/ Magazine etc referred.

Seminar Log Book Third Year Computer Engineering, SPPU, Pune Page 8
Annexure ii: Format for Seminar Report

Each student is required to write a comprehensive report about the seminar.

The report should be in the format as described below. It is important that
you adhere to these guidelines
A. Seminar report should be arranged as
1. Title Page with Title of the topic, Name of the candidate with Exam
Seat Number / Roll Number, Name of the Guide, Name of the
Department, Institution and Year & University
2. Seminar Approval Sheet/Certificate
3. Abstract and Keywords
4. Acknowledgements
5. Table of Contents, List of Figures, List of Tables and Nomenclature
6. Chapters Covering topic of discussion- Introduction with section
including organization of the report, Literature Survey/Details
of design/technology/Analytical and/or experimental work, if
any/
…..,Discussions and Conclusions ,Bibliography/References
7. Plagiarism Check report
8. Report Documentation page
B. Preparation Format
1) Report Size: Limit your seminar report to preferably 25- 40
pages.
2) Footer: The footer “Department of Computer Engineering,
PCCOER” should be included. It should be TIMES NEW ROMAN 10
pt. and right justified.
3) Header: The header “Seminar Title” centered and page
numbers on right should be included. Start numbering from
Introduction.
4) Paper Size : A- 4 size bond paper

5) Margins : Mirrored
1. Top : 1 inch
2. Bottom : 1 inch
3. Inside : 1.25 inch
4. Outside : 1 inch
6) Line Spacing: 1.5 lines
7) Title of Chapter
i. Font : Arial (Bold face, capital)
ii. Size : 16 point Alignment : centered
8) All Topics Headings
i. First Order Heading : (for example – 1.
INTRODUCTION)
1. Font : Times New Roman (Bold Face)
2. Size : 14 point
ii. Second Order Heading: (for example – 1.1.
Evolution)
1. Font : Times New Roman ( Bold Face)
2. Size : 12 point
Seminar Log Book Third Year Computer Engineering, SPPU, Pune Page 9
 iii. Third Order Heading: (for example – 1.1.1. Image
Processing)
1. Font : Times New Roman (Normal Face)
2. Size : 12 point
9) Text:
i. Font : Times New Roman
ii. Size : 12 point

10) Figures and Tables:

i. Caption: (for figures below the figure and for tables
above the table)
1. Font : Garamond (Bold)
2. Size : 11 point
3. Alignment : Center
11) References:
i. Book
Author name(s), Book Title, Publisher, Copyright Year, page
nos. if any.
ii. Journal/ Magazine/ Periodical
Author name(s), paper name, Journal/ Magazine/
Periodical name, issue no., page nos.
iii. Web Resources
Complete URL including File name.

Seminar Log Book Third Year Computer Engineering, SPPU, Pune Page 10
Annexure iii: Format for Cover Page

Seminar Report
On

Object detection for vehicles and computer components

By

Shashank Birajdar

[Exam No: 72160420L]

Under the guidance

Of

Mr. Mahendra Salunke

Department of Computer Engineering

Pimpri Chinchwad College of Engineering and Research,Ravet

SAVITRIBAI PHULE PUNE UNIVERSITY

2022 -2023

Seminar Log Book Third Year Computer Engineering, SPPU, Pune Page 11
Annexure iv.: Certificate

Department of Computer Engineering,

Pimpri Chinchwad College of Engineering and Research,Ravet

CERTIFICATE
This is to certify that from Third Year 2022-23

Engineering has successfully completed his / her seminar work titled ’’Object

detection for vehicles and computer components” at Pimpri Chinchwad

College of Engineering and Research , Ravet in the partial fulfillment of the

Bachelors Degree in Engineering.

Mr. Mahendra Salunke Mrs.Archana Chougule Mr.Harish Tiwari

Guide Head of the Department Principal
Seminar Log Book Third Year Computer Engineering, SPPU, Pune Page 12

Abstract

Visual surveillance for autonomous

power facility inspection is considered to be the
prominent field of study in the power industry. This
research field completely focuses on either object
detection or image fusion which lacks the overall
consideration. By considering this, a single end-to-
end object detection method by incorporating the
image fusion named Fast Detection Fusion
Network (FDFNet) is proposed in this paper to
output qualitative fused images with detection
results. The parameters in the FDFnet are greatly
reduced by sharing the feature extraction network
between image fusion and object detection tasks,
due to which a huge reduction in computational
complexity is achieved. On this basis, the object
detection algorithm performance on various types of
power facility images is compared and analyzed.
For experimentation purposes, an IR (infrared)
and VIS (visible) image acquisition system has
also been designed. In addition, a dataset named
CVPower with different sets of images for power
facility fusion detection is constructed for this
research field.
very large benchmarks for object
detection and viewpoint estimation. In this paper,
we tackle the problems of few-shot object detection
and few-shot viewpoint estimation. We demonstrate
on both tasks the benefits of guiding the network
prediction with class-representative features
extracted from data in different modalities: image
patches for object detection, and aligned 3D
models for viewpoint estimation. Despite its
simplicity, our method outperforms state-of-the-art
methods by a large margin on a range of datasets,
including PASCAL and COCO for few-shot object
detection, and Pascal3D+ and ObjectNet3D for
few-shot viewpoint estimation.
Furthermore, when the 3D model is not available,
we introduce a simple category-agnostic viewpoint
estimation method by exploiting geometrical
similarities and consistent pose labelling across
different classes.
 Acknowledgments

It gives us great pleasure in presenting the

seminar report on ‘Object Detection for vehicles
and computer components’.

I would like to take this opportunity to thank

my internal guide Prof. Mahendra Salunke for
giving me all the help and guidance I needed.
I am really grateful to them for their kind
support. Their valuable suggestions were
very helpful.

I am also grateful to Prof. Archana Chougule,

Head of Computer Engineering Department,
Pimpri Chinchwad College of engineering and
research for her indispensable support,
suggestions.

In the end our special thanks to Dr. H.U.

Tiwari for providing various resources such
as laboratory with all needed software
platforms, continuous Internet connection, for
our seminar.

Shashank Birajdar
(T.E. Computer
Engg.)
Contents

1 Synopsis 8
1.1 Seminar Title . . . . . . . . . . . . . . . . . 8
1.2 Internal Guide . . . . . . . . . . . . . . . . 8
1.3 Technical Keywords (As per ACM Key-
words) . . . . . . . . . . . . . . . . . . . . . 8
1.4 Problem Statement . . . . . . . . .
. . . . 9
1.5 Abstract . . . . . . . . . . . . . . . . . . . . 9
1.6 Goals and Objectives . . . . . . . . .
. . . 9
1.7 Relevant mathematics associated with
the Seminar
10
2 Technical Keywords 11
2.1 Area of Seminar ...................................... 11
2.2 Technical Keywords ............................... 11
3 Introduction 12
3.1 Seminar Idea ............................................ 12
3.2 Motivation of the Seminar ...................... 12
3.3 Introduction Part ....................................13
3.4 Literature Survey .................................... 13

4 Propose System 14

5 Requirements 15
5.1 Hardware Resources Required ............. 15
5.2 Software Resources Required ............... 15

6 Results 16
7 Summary and Conclusion 17

8 Plagiarism Report 18
List of Figures
List of Tables

3.1 Literature Survey ..................................... 13

5.1 Hardware Requirements ........................15
 Chapter 1Synopsis

1.1 Seminar Title

Object detection for vehicles and computer
components

1.2 Internal Guide

Prof. Mahendra Salunke

1.3 Technical Keywords (As per ACM

Keywords)
Please note ACM Keywords can be
found :
http://www.acm.org/about/class/ccs98-
html
Example is given as
1. C. IMAGE PROCESSING AND
COMPUTER VI- SION
(a) C.2 Digitization and Image Capture
i. C.2.4 Compression (Coding)
A. Enhancement
B. Restoration
C. Reconstruction
D. Feature Measurement
E. Scene Analysis
 F. Image Representation

1.4 Problem Statement

Object detection for vehicles and computer
components

1.5 Abstract
• With the development of artificial intelligence,
com- puter vision technology that simulates
human vi- sion has received widespread
attention. Based on the current commonly
used method of computer vi- sion technology-
deep learning, this paper outlines the
development of deep learning models, and de-
termines the inflection point of the
development of the introduction of
convolutional neural networks. Not only that,
convolutional neural networks com- bined with
deep learning have developed rapidly and are
widely used in the field of computer vision.
Therefore, the structure and development
process of the convolutional neural network
are further an- alyzed. Subsequently, the
three major difficulties in the field of computer
vision-image classification, object detection,
and image segmentation are com- pared and
analyzed, and their differences and con-
nection analysis are drawn. Finally, an
overview of the difficulties to be optimized in
computer vision, and computer vision
technology It will continue to develop as a
research hotspot.

1.6 Goals and Objectives

• 1. To provide computers the vision in every

aspect needed to analyze the objects like
human and pro- cess the situational
functionalities.
2. Identification of targeted object in moving
se- quence.
 3. To enable computing for particular
situations for targeted object movements.

1.7 Relevant mathematics associated with

the Seminar
System Description:
• Input:The MS COCO dataset is a large-scale
object detection, segmentation, and captioning
dataset pub- lished by Microsoft. Machine
Learning and Com- puter Vision engineers
popularly use the COCO dataset for various
computer vision projects.
• Output:Visualizing and exploring your
dataset.Generating predictions from an object
detection model.Evaluating mAP of a model on
your COCO dataset.
• For this experiment we placed randomly
50Cora) of the instances into the training set.
The rest (50and 75learned the pairwise
decision model on the training set.
Afterwards both classic and con- strained
versions of the HAC variations single, com-
plete and average linkage made an overall
decision of the whole dataset. The classic
versions use no constraints and only made
decisions based on the
trained pairwise decision model. So the classic
method corresponds to state-of-the-art object
identification without constraints. The
constrained versions are additionally guided
by the given constraints in the training set.
• Success Conditions: The input processing is
suc- cessful if the provided data is noise free
and prere- cognized by the system componetns
or the dataset and in normal condtions of
weather.
• Failure Conditions: The system can cause
failure in noisy data provided or, if the normal
condition such as weather is unsuitable for
devices.
 Chapter 2

Technical Keywords

Area of Seminar
Image processing

2.2 Technical Keywords

I. 4: IMAGE PROCESSING AND

COMPUTER VISION
I.4.0 : General
I.4.1 : Digitization and Image
Capture I.4.2: Compression
(Coding)
I.4.3:
Enhancement
I.4.4: Restoration
I.4.5:
Reconstruction
I.4.6:
Segmentation
I.4.7: Feature
Measurement I.4.8:
Scene Analysis I.4.9:
Applications
I.4.10: Image Representation
I.4.m: Miscellaneous
 Chapter 3 Introduction

Computer vision is a field of artificial intelligence (AI) that enables

computers and systems to derive meaningful information from digital
images, videos and other visual inputs.While a convolutional neural
network understands single images, a recurrent neural network
processes video inputs to enable comput- ers to ‘learn’ how a series of
pictures relate to each other. In 2015, technology leader Google rolled
out its instant translation service that leverages com- puter vision
through smartphone cameras. Neural Machine Translation, a key
system that drives instantaneous and accurate computer vision-based
translation, was incorporated into Google Translate web results in
2016. Not to be left behind, technology giant Meta (earlier known as
Facebook) is also dabbling in computer vision for various exciting
applications.

3.1 Seminar Idea

Computer vision as a discipline is at the cutting edge of science. As
with any frontier, it is thrilling and chaotic, with often no trustworthy
authority to turn to. Numerous beneficial concepts lack a theoretical
foundation, and some theories are rendered ineffective in reality;
developed regions are widely dispersed, and often one seems totally
unreachable from the other

3.2 Motivation of the Seminar

The motive of object detection is to recognize and locate

(localize) all known objects in a scene.Preferably in 3D space,
recovering pose of objects in 3D is very important for robotic control
systems. The
 information from the object detector can be used for obstacle
avoidance and other interactions with the environment.

3.3 Introduction Part

• In many autonomous driving systems, the object detection subtask
is it- self one of the most important prerequisites to autonomous
navigation, as this task is what allows the car controller to account
for obstacles when considering possible future trajectories; it
therefore follows that we desire object detection algorithms that are
as accurate as possible.
Car computing systems are often heavily memory constrained, and
so it is often infeasible to store and run detectors with large
amounts of pa- rameters, especially with large input image
volumes. In particular, this constrains the depth of neural network
approaches.YOLO and RCNN is a recent model that operates
directly on images while treating object detection as regression
instead of classification. YOLO has the poorest performance out
of all the above models, but more than makes up for it in speed;
YOLO is able to predict at an astounding 45 frames per second.
YOLO is, however, quite large at 12 layers.

3.4 Literature Survey

S Author name P
r.
Xiang Xu An End to End Object Detection Based on Heterogeneous
Image Fusion
Z. Chen, Road lane detection
R.
(Dulari Bhatt intelligent driving and security monitoring
1
Steffen Object Identification with Constraints height
Rendle
Table 3.1: Literature Survey
Chapter 4 Propose System

The R-CNN consists of 3 main modules:

1. The first module generates 2,000 region proposals using the Selective
Search algorithm. After being resized to a fixed pre-defined size
2. the second module extracts a feature vector of length 4,096 from each
region proposal.
3. The third module uses a pre-trained SVM algorithm to classify the region
proposal to either the background or one of the object classes.
Fast R-CNN Quick Overview: instead of trying to classify a huge number
of regions, you can just work with 2000 regions. These 2000 region
proposals are generated using the selective search algorithm. These 2000
candidate region proposals are warped into a square and fed into a
convolutional neural network that produces a 4096-dimensional feature
vector as output. The CNN acts as a feature extractor and the output
dense layer consists of the features extracted from the image and the
extracted features are fed into an SVM to classify the presence of the
object within that candidate region proposal.
The authors emphasize that YOLOS is, at this moment, a proof-of-
concept with no performance optimizations, its purpose was to prove
the Transform- ers’ versatility and transferability to the more
challenging object detection task, than just image recognition.
YOLOS consists of a series of object detection models based on
the ViT architecture with the fewest possible modifications and
inductive biases.
For the pre-training dataset was used the mid-sized ImageNet-1, and the
vanilla ViT architecture, with the fewest possible modifications. For the first
time, the 2D object detection was performed in a pure sequence-to-
sequence manner by taking a sequence of fixed-sized non-overlapping
image patches.
 Chapter 5 Requirements

5.1 Hardware Resources Required

Sr. Parameter Minimum Justification

No. Requirement
1 CPU 2 GHz Remark
Speed Required
2 RAM 3 GB Remark
Required
Table 5.1: Hardware Requirements

5.2 Software Resources Required

Platform :
1. Operating System: windows\linux
2. IDE ,Programming Language: opencv.
Chapter 6Results

Provided the computer a vision to analyze the objects and human

beings. Identification of targeted object in moving sequence. Detection of
moving ob- jects around the car and alerting the driver, this technology
enhances safety when pulling out of a parking space, helping give
drivers better awareness of their surroundings and increased peace of
mind.
Chapter 7

Summary and Conclusion

object detection is problem of detecting the location of individuals who are

walking on a particular indoor and outdoor environment or objects. It
is an essential and significant task in any intelligent video surveillance
system, as it provides the fundamental information for semantic
understanding of the video footages.object detection is problem of
detecting the location of individuals who are walking on a particular
indoor and outdoor environment or objects. It is an essential and
significant task in any intelligent video surveillance system, as it provides
the fundamental information for seman- tic understanding of the video
footages. Two essential technologies are used to accomplish this:
machine learning and a convolutional neural network (CNN). A recurrent
neural network (RNN) is used in a similar way for video applications to
help computers understand how pictures in a series of frames are related
to one another.MS COCO (Microsoft Common Objects in Con- text) is a
large-scale image dataset containing 328,000 images of everyday objects
and humans. The dataset contains annotations you can use to train
machine learning models to recognize, label, and describe objects.In yolo,
continuous 3000 frames of images were used for vehicle detection in a
variety of highway scenes by using our trained model. We extracted and
divided the road surface area and put it into the network for vehicle
detection. Then, we compared our method with the detection of images
with 1920*1080 resolution into the network (without dividing the road
surface).In many autonomous driving systems, the object detection
subtask is itself one of the most im- portant prerequisites to autonomous
navigation, as this task is what allows the car controller to account for
obstacles when considering possible future trajectories; it therefore
follows that we desire object detection algorithms that are as accurate
as possible. Many high-quality object detectors have seen astounding
improvements in recent years.
Chapter 8 Plagiarism Report

Plagiarism report
 REFERENCES.
[1] Gene Lewis Stanford University Stanford, CA:
Object Detection for Autonomous Vehicles
[2] Xiang Xu, Gang Liu, Member, IEEE, Durga
Prasad Bavirisetti, Xiangbo Zhang, Student
Member, IEEE, Boyang Sun, and Gang Xiao: Fast
Detec- tion Fusion Network (FDFnet): An End to
End Object Detection Framework Based on
Heterogeneous Image Fusion for Power Facility
Inspection [2] Ab- hishek Balasubramaniam,
Sudeep Pasricha:Object Detection in Autonomous
Vehicles: Status and Open Challenges
[3] Teutsch Michael: Moving Object Detection
and Segmentation for Re- mote Aerial Video
Surveillance.
[4] https://www.ibm.com/in-
en/topics/computer-vision [5]
https://ijcrt.org/papers
Annexure v: Report Documentation
Seminar Report Documentation

Report Code: CS-TE-Seminar 2022-2023 Report Number: TYCOB-03

Report Title: Object detection for vehicles and computer components.

Address (Details):
Pimpri Chinchwad College of Engineering and Research , Ravet.
Author [with Address, phone, E-mail]:8956634575.
Shashank.birajdar_comp2020@pccoer.in

E-mail :shashank.birajdar_comp2020@pccoer.in
Roll: TYCOB-03

Cell No:8956634575

Year: 2022 – 2023

Branch: Computer Engineering
Key Words:Object Detection.

Type of Report Guides Complete

Report: Report Checked Name: Total
FINAL Checked By: Date: Copies
Mr. mahendra
Salunke 2

Abstract:
Modern vehicles are equipped with the different systems that help
driver in the driving process ensuring safer and more comfortable
driving. These systems are called Advanced Driver Assistance
Systems (ADAS) and are step toward fully autonomous driving. The
integral part of autonomous driving is an object detection and tracking
by using front view camera which provides necessary information for
emergency braking, collision avoidance, path planning, etc. In this
paper, one possible approach to object detection and tracking in
autonomous driving is presented. Two object detection methods are
implemented and tested: Viola-Jones algorithm and YOLOv3.

Seminar Log Book Third Year Computer Engineering, SPPU, Pune Page 13