CU6051NI

Module Code & Module Title

CU6051NI - Artificial Intelligence

Assessment Weightage & Type

75% Individual Coursework

Year and Semester

2023-24 Autumn

Student Name: Aarohan Subedi

London Met ID: 22015633
College ID: np01cp4s220149
Assignment Due Date:1/17/2024
Assignment Submission Date:1/17/2024

I confirm that I understand my coursework needs to be submitted online via Google Classroom under the
relevant module page before the deadline in order for my assignment to be accepted and marked. I am
fully aware that late submissions will be treated as non-submission and a marks of zero will be awarded
CU6051NI

Table of Contents
1. INTRODUCTION...................................................................................................... 1
1.1. AI Concept Used ................................................................................................ 1
1.2. Problem Domain ................................................................................................ 1
2. BACKGROUND ....................................................................................................... 2
2.1. Research Work Done on FRAS (Facial Recognition Attendance System) ......... 2
2.1.1. AI in a Vague Part of Facial Recognition Attendance System. .................... 2
2.1.2. AI in specific Part of the project (FRAS) ...................................................... 5
2.1.3. Advantages Of Facial Recognition Attendance System. ............................. 6
2.1.4. Disadvantages Of Facial Recognition Attendance System. ......................... 7
2.1.5. Dataset ........................................................................................................ 8
2.2. Review and Analysis of existing work ................................................................ 9
2.2.1. Review Of Existing System........................................................................ 10
2.2.2. Summarized Review and analysis. ............................................................ 11
3. SOLUTION ............................................................................................................. 12
3.1. Algorithm Of the System. ................................................................................. 12
3.2. Pseudo-Code. .................................................................................................. 13
3.3. Logical Flow chart. ........................................................................................... 14
3.4. State Transition Diagram. ................................................................................ 15
3.5. Development Process. ..................................................................................... 15
3.5.1. Tools Used ................................................................................................ 15
3.5.2. Tool Kit Used ............................................................................................. 17
3.5.3. Python Libraries Used. .............................................................................. 18
3.5.4. Explanation of development process. ........................................................ 18
3.5.5. Achieved Results. ...................................................................................... 21
4. CONCLUSION ....................................................................................................... 24
Bibliography .................................................................................................................. 25
APPENDIX ...................................................................... Error! Bookmark not defined.
CU6051NI

Table of Figures
Figure 1 Flow chart........................................................................................................ 14
Figure 2 System architecture diagram........................................................................... 15
Figure 3 Installing required packages............................................................................ 18
Figure 4 Importing required libraries.............................................................................. 18
Figure 5 Loading and encoding photos. ........................................................................ 18
Figure 6 creating array for known face and their names. .............................................. 19
Figure 7 Setting webcam to default. .............................................................................. 19
Figure 8 Initiating loop for real time face recognition and capturing video frame. .......... 19
Figure 9 detecting face location and encoding faces in video frame. ............................ 19
Figure 10 Matching faces and handling matches. ......................................................... 19
Figure 11 Writing to csv file. .......................................................................................... 19
Figure 12 Drawing rectangle and labelling faces. .......................................................... 20
Figure 13 Displays resulting frames. ............................................................................. 20
Figure 14 Existing the loop. ........................................................................................... 20
Figure 15 Releasing resource and closing window. ...................................................... 20
Figure 16 Detected unknown undefined faces. ............................................................. 21
Figure 17 successful face detection. ............................................................................. 22
Figure 18 Generated Csv with known faces date and attendance time......................... 23
CU6051NI

Table of tables

Table 1 Summarized table of different similar system. .................................................. 11

CU6051NI

1. INTRODUCTION

In the Field of Artificial Intelligence, Combination of Machine learning and computer vision
has made a way for innovative solutions to everyday challenges. For this coursework I
am making an AI project named “Facial Recognition Attendance System”(FRAS). This
system will be able to successfully recognize the face of a user saved in the database
and will be able to record the name and date of the user.

As the traditional attendance method is susceptible to inaccuracies, the implementation

of facial recognition will play as an alternative offering efficiency and precision.

1.1. AI Concept Used

The concept used in this project is Machine Learning. Machine learning is a subset of AI
which plays a main role in this project. Supervised Learning, i.e. a process where the
system learns patterns from the labeled data, is the main point to process of facial
recognition.

1.2. Problem Domain

When there are large number of students in university, it becomes hard to list the
attendance of the student and it consumes time. Traditional ideas to mark attendance are
proving inaccurate and are continuously facing problems. The old methods like calling out
student’s names and verifying ID-cards of students is a hassle. Especially in the lecture
class where there are large number of students old techniques fails. Thus, the face
recognition attendance system is proposed to overtake the manual attendance methods.
Furthermore, The Facial recognition attendance system will be able to overcome the fraud
approach and lecturers will not have to make sure the number of students presence in
the class several times by counting the name list.

1
22015633 Aarohan Subedi
CU6051NI

2. BACKGROUND

The earliest pioneers of facial recognition were Woody Bledsoe, Helen Chan Wolf, and
Charles Bisson. In 1964 and 1965, Bledsoe, along with Wolf and Bisson began work using
computers to recognize the human face. Due to the funding of the project originating from
an unnamed intelligence agency, much of their work was never published. However, it
was later revealed that their initial work involved the manual marking of various
“landmarks” on the face such as eye centers, mouth etc. These were then mathematically
rotated by a computer to compensate for pose variation. The distances between
landmarks were also automatically computed and compared between images to
determine identity. These earliest steps into Facial Recognition by Bledsoe, Wolf and
Bisson were severely hampered by the technology of the era, but it remains an important
first step in proving that Facial Recognition was a viable biometric (G.Robinson-Riegler,
2008).

2.1. Research Work Done on FRAS (Facial Recognition Attendance

System)
Detailed research is carried out on the topic. All the Research done are included as per
the subheading defined in this section.
2.1.1. AI in a Vague Part of Facial Recognition Attendance System.

1. Workplace & Educational Institute Attendance Tracking:

The application of facial recognition system in workplace eases the employee

clock in and out procedure saving the time along with less errors. The use of it
maintains attendance accuracy and its record report for the payroll and
performance analysis.

Facial recognition for the student attendance in educational institution reduces

the administrative burdens as the system monitors student attendance in
classrooms. The system helps to track and identify attendance patterns of the
students.

2
22015633 Aarohan Subedi
CU6051NI

• Trends: The system integrates with the workplace environment to handle the
employee information and automation for their payroll. The aim is to record
the details for coordination with the HR system.
Coordinating educational fields with facial recognition results as automated
attendance reports and gives a path to analyze the attendance patterns for
the management systems.

• Approach: Liveness detection such as blink detection, eye movement, 3D

depth sensing and challenge response protocols ensures that the data being
analyzed enhances the security along with real time user presence.

2. Security and Surveillance:

The use of facial recognition for security purposes is high these days. It is used
to control access over secured areas as it verifies the identity of an authorized
individual. This system is used even in law enforcement fields for criminal
identifications as they can match faces in surveillance recording with the criminal
databases.

• Trend: The Convolutional Neural Networks (CNNs) which are special

computer programs are used in finding details and recognizing the faces in it.
Amending small changes or adjustment with pre-trained models such as Arc
Face or Face Net these programs work better meeting the exact requirement
smoothly.

• Approach: The advanced computer programs which are specially using deep
learning are helping computers recognize faces more easily and accurately.
Also, these computer programs have taken advancement in figuring out if a
face is real, helping to stop fake faces tricking the system.

3. Automotive Industry:

In the field of automotive industry, facial recognition can be used for the
convenient and secure access of vehicles. The key based entry can be replaced
by facial recognition system. Further, it can also be used for the safety purpose
of monitoring driver’s facial features while driving.

• Trends: The Driver Monitoring System (DNS) level up in the automotive

industry incorporating facial recognition which monitors the driver’s facial
expressions, attention intensifying the safety. It also sums up access control
and provides security where it prevents unauthorized access to the vehicle.

3
22015633 Aarohan Subedi
CU6051NI

• Approach: The live detections such as eye movements, blinking of eye, facial
expressions, 3D depth sensing along with challenge respond protocol is the
approach that covers the technical and ethical aspect while implementing the
facial recognition system.

4. Individual’s Authentication and Authorization:

Facial recognition can be applied to any individual to log in into their devices. The
system authenticates and authorizes the individual making them an ease without
the need for passwords. Certain platforms use facial recognition for secure
authentication ensuring the user’s account/device’s security.

• Trends: The multi-factor authentication which includes the combination of

biometrics (iris scans, fingerprints) running with facial recognition can
enhance security, known to be biometric fusion. While implementing such,
privacy concern arises thus maintaining biometric data encryption is
necessary. Hence, privacy preserving technologies must be used to maintain
privacy while still being effective.

• Approach: Federated learning where the models collaborate without sharing

any of the user’s data and maintains the privacy concerns is approached.
Measures like differential privacy, secure enclaves and anonymization
techniques are used to keep the sensitive facial data safe.

5. Healthcare:

In the field of healthcare, facial recognition can identify the patient on the
database and can assist in providing the details while monitoring and recording
the patient’s status. It helps reduce the risk of errors and maintains the accuracy
of timely conditions while monitoring the patient.

• Trends: After the pandemic, there has been a rise in contactless

authentication for minimizing physical contact in the healthcare field. Thus,
the identification and access of patient’s details is done via facial
recognition system. It also reduces the administrative paperwork as the
system automates routine tasks.
• Approach: The active Learning approach is widely used in this sector as
the system can learn from the feedback provided maintaining the data. It
also collaborates with federated learning which shows the approach of
allowing the models without sharing the data of any user.

4
22015633 Aarohan Subedi
CU6051NI

2.1.2. AI in specific Part of the project (FRAS)

Facial recognition attendance systems are powerful applications that leverage computer
vision technology to automate and streamline the attendance tracking process. The
system uses a camera to capture real-time video images, extract facial features from it,
and uses advanced algorithms to generate a facial encoding. By comparing these facial
coding with pre-registered facial coding, the system identifies a person and marks their
presence. During operation, the system continuously captures individual images,
recognizes faces, and encodes facial features for comparison to a database of known
faces. When a face is detected, the system associates it with the corresponding ID and
records the presence information in a CSV file. CSV files typically contain important
details such as the person's name, date, and time of attendance, etc. This application has
several benefits, primarily improving accuracy, efficiency, and security of time and
attendance management. Traditional methods such as manual time and attendance
management and card-based systems can be error-prone and time-consuming. Facial
recognition eliminates the need for physical contact or additional identification methods,
ensuring a contactless and seamless experience. Additionally, system automation
reduces administrative burden and allows attendance data to be tracked and reported in
real-time. Additionally, the generated CSV file provides a convenient and organized time
and attendance record, facilitates integration with other systems, and simplifies data
analysis. Additionally, facial recognition attendance systems enhance security by
ensuring that attendance records match individuals based on unique facial
characteristics, thereby minimizing the risk of fraudulent attendance. This application is
widely used in various educational institutions, corporate offices, and other organizations
where accurate attendance tracking is essential. Overall, it represents a modern and
efficient time and attendance solution that leverages the power of facial recognition
technology to provide a reliable and user-friendly experience.

5
22015633 Aarohan Subedi
CU6051NI

2.1.3. Advantages Of Facial Recognition Attendance System.

The implementation of the Facial recognition attendance system has multiple

advantages, showing its implementation in various fields. Some of the key positive
aspects are:
• Productivity and Accuracy:
Saves time with computerized attendance tracking.
Guarantees high precision in recognizing people.
• Simplified Administrative Processes:
Computerized record-keeping simplifies authoritative tasks.
Effortless generation of comprehensive attendance reports.
• Improved Security:
Includes an additional layer of security through identity confirmation.
Decreases the hazard of false participation.
• Cost-Effective:
Diminishes administrative costs related with manual tracking.
• Contactless Solution:
Gives a sterile and secure alternative to traditional strategies.
• Versatility:
Versatile to different settings, making it flexible.
• User-Friendly Involvement:
Offers a helpful attendance tracking process for clients.
• Innovative Progressions:
Exhibits development and commitment to advanced solutions.
• Data Insights:
Allows examination for attendance pattern insights.
• Compliance and Accountability:
Makes review trails, improving accountability and consistency.

6
22015633 Aarohan Subedi
CU6051NI

2.1.4. Disadvantages Of Facial Recognition Attendance System.

Addressing the challenges below is essential for the responsible development and
deployment of Facial recognition attendance system.

• Privacy Concerns:
Facial recognition attendance systems include the collection and handling of delicate
biometric information, driving to concerns around the potential attack of individuals'
security.
The storage of facial information in frameworks like CSV records raises the hazard of
unauthorized access and abuse, requiring strong security measures.
• Precision Challenges:
Inconstancy in lighting conditions, facial expressions, and picture quality within the
dataset can result in mistakes, driving to false positives or negatives during
identification.
Biases within the training data may contribute to unfair results, affecting certain
statistical groups and raising moral concerns.
• Technical Restrictions:
Executing a viable facial recognition system may require high-quality cameras and
critical handling capabilities, posturing potential monetary limitations for organizations
with restricted assets.
Continuous support and upgrades are vital to adjust to advancing innovative guidelines
and upgrade security measures.
• Legal and Ethical Issues:
Compliance with information security regulations, such as GDPR, is significant,
requesting meticulous consideration to consent, transparency, and secure information
handling.
Ethical considerations include issues of mass observation, individual consent, and the
mindful use of facial recognition technology, emphasizing the need for a adjusted and
moral approach.

7
22015633 Aarohan Subedi
CU6051NI

2.1.5. Dataset

The choice of datasets is an essential decision within the improvement of a Facial

Recognition Attendance System (FRAS) that focuses on pictures of students in a class.
The explanation for selecting this dataset is rooted in its significance, characteristics,
and its significant contribution to addressing the distinguished issue.

• Relevance to Attendance Tracking:

The dataset comprises of pictures of students, specifically aligning with the objective of
attendance tracking in an educational system. This specificity guarantees that the FRAS
is prepared on significant facial highlights, optimizing its execution for recognizing and
enlisting students' attendance.

• Inherent Class Dynamics:

By utilizing pictures of students inside the class, the dataset covers the inherent
elements of the learning environment. This includes varieties in student appearances,
expressions, and postures that are typical amid regular class exercises, contributing to
the FRAS's flexibility to real-world scenarios.

• Demographic Representation:
The dataset is carefully curated to include a differing representation of student
demographics, considering variables such as sex, ethnicity, and age. These differences
improve the FRAS's capacity to recognize and separate among a wide range of student
faces, promoting inclusivity.

• Consistency in Picture Source:

Pictures sourced particularly from the student body inside the class guarantee
consistency within the facial information. This consistency is useful for preparing the
FRAS to handle pictures captured inside the same environment, minimizing
inconstancy, and improving precision.

8
22015633 Aarohan Subedi
CU6051NI

• Ethical Considerations and Consent:

The use of student pictures requires a solid emphasis on ethical contemplations and
getting appropriate consent. Guaranteeing that the dataset follows privacy controls and
ethical benchmarks is essential to building belief among students.

• Practical Applicability:
The dataset's focus on students inside a class specifically addresses the practical issue
of attendance tracking in an educational system. This significance upgrades the FRAS's
applicability and viability within the setting for which it is designed.

• Flexibility to Classroom Conditions:

The dataset incorporates pictures captured beneath changing classroom conditions,
accounting for diverse lighting scenarios and potential challenges related to student
positioning. This versatility guarantees that the FRAS is robust in recognizing faces
within the dynamic environment of a classroom.

• Contribution to System Accuracy:

The carefully chosen dataset plays a essential part in preparing the FRAS to attain high
precision in facial recognition. The significance and characteristics of the dataset
directly contribute to the system's viability in precisely recognizing and recording student
attendance.

2.2. Review and Analysis of existing work

The review and analysis of the system related to Facial recognition attendance system
is necessary to make sure the system I am trying to make is free from the mistakes
done by the previous developers for smooth running and deployment of the system.

9
22015633 Aarohan Subedi
CU6051NI

2.2.1. Review Of Existing System.

• Online Attendance system based on facial recognition.

Aurther Name: T. Sharaniya, Year: 2020 (T.Sharanya, 2020).
There are many biometric processes out of which face recognition is the best method.
In this method the camera is fixed within the classroom, and it’ll take the image, the
faces are detected and are recognized with our database and finally the attendance is
listed.

• Design and evaluation of a face recognition system using CNN.

Aurther Name: Pranav K.B, Year: 2020 (PranavK.B, 2020).

Face Recognition system uses offline data or real time input, based on the application
of the project. In this research a real time face recognition system is made using CNN
(Convolutional Neural Network) is proposed.

• Electronic Attention System using CNN and Open cv.

Aurther Name: Omkar Sawant, Jash Jain. Year: 2019 (Omkar Sawant, 2019).
This system automatically detects students when they enter the classroom and takes
the attendance by recognition and absentees list are displayed in the app.

• Large Pose face alignment through CNN based dense 3D Model Fitting.

Author name: Amin Jourabloo, Xiaoming Liu. Year: 2016 (Amin Jourabloo,
2016).

They proposed a method to fit a 3D dense shape to a face image with large poses by
combining cascade CNN repressors and the 3D Morph Able Model (3DMM). They put
forward two types of pose invariant features for boosting the precision of face alignment.

10
22015633 Aarohan Subedi
CU6051NI

• Face Recognition system using Principal Component Analysis, Eigen face

and Neural Network.
Author Name: A. S. SYED NAVAZ1, T. DHEVI SRI2 & PRATAP MAZUMDER
Year:2013. (A. S. SYED NAVAZ1, 2013).

The research paper shows a face recognition approach using PCA and Neural Network
techniques. The obtained result is compared with K-means, Fuzzy Ant with fuzzy C-
means and presented technique provides a better recognition rate then the other two.

2.2.2. Summarized Review and analysis.

Table 1 Summarized table of different similar system.

SN Title Concept Used Advantages Disadvantages

1 Individual Face recognition It projects on face Multiple
Stable Space: under uncontrolled recognition under persons per
An Approach condition. uncontrolled image is
to Face conditions. required.
Recognition
Under
Uncontrolled
2 Anti-Cheating Dual vision face It will anticipate Angle of user
Presence recognition using face data with face will affect
System Based 3WPCA. recognition the accuracy.
on 3WPCA accuracy to 90%.
Dual Vision
Face
Recognition.
3 Prototype ADA Boost This system Only valid for a
model for an algorithm with reduces the single image of
Intelligent techniques chances of fake a system.
Attendance PCA and LDA attendance and
System based. Hybrid algorithm proxy.
on facial
Identification.
4 Automated DWT Facial images can Higher
Attendance be recognized accuracy can
System Using successfully be obtained.
Face
11
22015633 Aarohan Subedi
CU6051NI

Recognition where recognition

is 82%.
5 An Attendance MATLAB User friendly Similar used
Marking System techniques.
based on Face
Recognition.
6 Face recognition n- Local Binary Continuous and System has
based Attendance patterns (Support automatic issues with
System using vector machine), attendance system
Machine Learning LDA based OpenCV system. performance
Algorithms Radhika and FLTK and accuracy.
C. Damale

3. SOLUTION
3.1. Algorithm Of the System.

The code employs the given face recognition library, which is built on top of the dlib
library. The fundamental face recognition algorithm utilized by dlib could be a
combination of profound learning and classic picture preparing strategies. The key
calculation for face recognition is the confront encoding calculation, which changes
over a face picture into a numerical representation, or encoding, that can be compared
to other face encodings to decide if they match.

Here is the breakdown to face recognition algorithm used in the project.

Face Detection: The face_recognition. face_locations function is used to detect face
loca in the captured images. This function create a (HOG) based detection, which is a
classic computer vision technique.

Face Encoding: The face_recognition.face_encodings function brings out the facial

features and feeds them into a numerical values. This process contains a deep neural
network that has already been trained on a database.

12
22015633 Aarohan Subedi
CU6051NI

Face Comparison: The code uses the face_recognition.compare_faces function to

compare the captured face with the known given face . This function makes sure if a
given face coding corresponds to a known face.

Attendance Log: If the match is found the script logs the attendance by writing the
name of the known student and the current time to a new CSV file.

3.2. Pseudo-Code.

IMPORT face_recognition as fr
IMPORT cv2
IMPORT csv
FROM date time IMPORT datetime
LOAD the known images and their encodings
OPEN the video capture device
WHILE the video is running:
READ the next frame
FIND all faces in the frame
FOR each face:
COMPARE the face to the known face encodings
IF a match is found, write the name of the person to a CSV file
DRAW a rectangle around the face
WRITE the name of the person on the face
DISPLAY the frame
IF the `q` key is pressed, stop the video
RELEASE the video capture device
END

13
22015633 Aarohan Subedi
CU6051NI

3.3. Logical Flow chart.

Figure 1 Flow chart

14
22015633 Aarohan Subedi
CU6051NI

3.4. State Transition Diagram.

Figure 2 System architecture diagram

3.5. Development Process.

3.5.1. Tools Used

Python:
Python serves as the essential programming dialect in Python Idle, advertising a flexible
and expressive sentence structure. Famous for its straightforwardness and coherence,
Python empowers consistent improvement of the confront acknowledgment framework.
Its broad library back and community commitments make it an perfect choice for
actualizing machine learning and computer vision calculations.

OpenCV:
OpenCV, consistently coordinates into Python Sit still, could be a effective computer
vision library. It prepares the extend with comprehensive picture preparing
functionalities basic for confront location and acknowledgment. OpenCV's feature-rich
toolkit incorporates apparatuses for picture control, protest following, and design
acknowledgment, making it crucial for extricating significant data from visual
information.

15
22015633 Aarohan Subedi
CU6051NI

NumPy:
NumPy, congruous with Python Sit out of gear, may be a essential library for numerical
computing. Its cluster operations and mathematical capacities improve the effectiveness
of information control, basic for preparing picture information within the venture.
NumPy's multidimensional clusters encourage complex computations, advertising a
strong establishment for dealing with and analyzing numerical information in a brief and
expressive way.

Face_recognition Library:
The face_recognition library, consistently coordinates with Python Sit still, disentangles
complex confront acknowledgment errands. Leveraging pre-trained models, it
streamlines the method of confront discovery and acknowledgment. By abstracting
complicated calculations, face_recognition permits engineers to center on the
application's high-level rationale, upgrading efficiency and diminishing the complexity
related with actualizing facial acknowledgment highlights.
Dlib:
Dlib, interfaces with Python Sit still, could be a C++ library known for its productivity in
machine learning and computer vision errands. Within the face acknowledgment
extend, Dlib's part is vital for facial point of interest discovery, contributing to the exact
recognizable proof of key facial highlights. Its strong facial investigation capabilities
make it an important device for improving the precision and unwavering quality of the
confront acknowledgment framework.

CV2 (OpenCV for Python):

CV2, a Python IDLE-compatible OpenCV wrapper, extends the capabilities of OpenCV
inside the Python environment. It encourages a wide range of computer vision
highlights, counting capturing video outlines and creating graphical client interfacing.
CV2 upgrades real-time video handling and visualization, guaranteeing the project's
flexibility to differing applications, from video reconnaissance to intelligently client
interfacing.

Python Idle:
Python Idle serves as the essential improvement environment, giving an natural
interface for composing, testing, and executing Python code. Its simplicity makes it an
open choice for designers, particularly those modern to programming. Python IDLE's
16
22015633 Aarohan Subedi
CU6051NI

intuitively shell encourages speedy experimentation, permitting designers to iteratively

refine their code and troubleshoot issues, contributing to a smooth and effective
advancement workflow.
IDEs:
Coordinates Advancement Situations (IDEs) like PyCharm, congruous with Python Sit
out of gear, upgrade the development process with progressed instruments. These
incorporate modern code editors, coordinates investigating, and venture administration
features. PyCharm, in specific, offers a user-friendly interface, code completion, and
version control integration, giving a comprehensive advancement environment that
streamlines coding errands and cultivates collaboration among engineers.
These nitty gritty clarifications emphasize the noteworthiness of each apparatus within
the confront acknowledgment venture, emphasizing their one-of-a-kind commitments
and collaborative part in accomplishing a strong and effective facial acknowledgment
framework.
3.5.2. Tool Kit Used

Dlib:
Task:
Dlib serves as a foundational C++ library with effective instruments for machine learning
and computer vision. Within the confront acknowledgment extend, Dlib's essential part
is facial point of interest location, contributing to the exact distinguishing proof of facial
highlights.
Importance:
Dlib's vigorous capabilities in facial investigation improve the exactness and unwavering
quality of the confront acknowledgment framework. Its productive execution guarantees
the project's capacity to distinguish key facial landmarks, a basic step within the in
general acknowledgment handle.

face_recognition Framework:
Task:
The face_recognition system disentangles complex confront acknowledgment
assignments by leveraging pre-trained models for confront location and
acknowledgment.

17
22015633 Aarohan Subedi
CU6051NI

Importance:
By abstracting complex calculations, face_recognition permits for a higher-level usage
of facial acknowledgment highlights. This contributes to expanded efficiency,
decreasing the complexity related with the usage of facial acknowledgment
assignments.

3.5.3. Python Libraries Used.

• Face_recognition: It is an ai model that scans and recognize human faces.

• Cv2: It is opencv-python package.
• Csv: It will be used to manipulate data in csv file.
• Os: It handles files and folders.
• NumPy: It is used for numerical calculations.

3.5.4. Explanation of development process.

Figure 3 Installing required packages.

Figure 4 Importing required libraries.

Figure 5 Loading and encoding photos.

18
22015633 Aarohan Subedi
CU6051NI

Figure 6 creating array for known face and their names.

Figure 7 Setting webcam to default.

Figure 8 Initiating loop for real time face recognition and capturing video frame.

Figure 9 detecting face location and encoding faces in video frame.

Figure 10 Matching faces and handling matches.

Figure 11 Writing to csv file.

19
22015633 Aarohan Subedi
CU6051NI

Figure 12 Drawing rectangle and labelling faces.

Figure 13 Displays resulting frames.

Figure 14 Existing the loop.

Figure 15 Releasing resource and closing window.

20
22015633 Aarohan Subedi
CU6051NI

3.5.5. Achieved Results.

Figure 16 Detected unknown undefined faces.

21
22015633 Aarohan Subedi
CU6051NI

Figure 17 successful face detection.

22
22015633 Aarohan Subedi
CU6051NI

Figure 18 Generated Csv with known faces date and attendance time.

23
22015633 Aarohan Subedi
CU6051NI

4. CONCLUSION

Analyzing the work done in the development of facial recognition attendance systems
reveals the effectiveness and importance of the research. This system satisfies the
objectives outlined in the introduction and provides an efficient and accurate way to
automate attendance tracking. Using facial recognition technology, the system matches
her face with known identities and creates a CSV file for her with relevant attendance
details. In addition to achieving its technical goals, this project demonstrates the
practicality of integrating facial recognition into time and attendance systems and
demonstrates the accuracy and speed of real-time recognition.

This solution effectively addresses real-world problems related to attendance tracking. In

educational institutions, businesses and various organizations, this system offers an
optimized alternative to traditional methods. This reduces administrative burden,
improves workflow, and facilitates contactless time and attendance management that
meets hygiene practice requirements. The system's practical application has spread
across various industries, providing an easy-to-use and reliable time and attendance
solution that contributes to overall efficiency and security.

However, it is important to recognize the limitations of facial recognition attendance

systems. Performance can be affected by factors such as lighting conditions, camera
quality, and facial image clarity. If people wear masks or change their appearance
significantly, the system's accuracy can be affected. Privacy and data security concerns
also need to be carefully considered, highlighting the importance of balancing
technological innovation and ethical considerations.

Future improvements may focus on improving detection accuracy under different

conditions and incorporating additional safety measures. Continued research and
development are required to address these limitations and ensure the system's
adaptability to evolving real-world challenges. Finding a balance between technological
innovation and ethical considerations continues to be important for the successful
implementation and implementation of facial recognition solutions.
24
22015633 Aarohan Subedi
CU6051NI

Bibliography
A. S. SYED NAVAZ1, T. D. S. &. P. M., 2013. researchgate. [Online]
Available at:
235950165_FACE_RECOGNITION_USING_PRINCIPAL_COMPONENT_ANALYSIS_
AND_NEURAL_NETWORKS
Amin Jourabloo, X. L., 2016. CVF Open Access. [Online]
Available at:
https://openaccess.thecvf.com/content_cvpr_2016/papers/Jourabloo_Large-
Pose_Face_Alignment_CVPR_2016_paper.pdf
Anaconda.com, 2023. anacondanevigator. [Online]
Available at:
https://docs.anaconda.com/free/navigator/index.html#:~:text=Anaconda%20Navigator%
20is%20a%20desktop,line%20interface%20(CLI)%20commands.
G.Robinson-Riegler, 2008. NEC. [Online]
Available at: https://www.nec.co.nz/market-./publications-media/a-brief-history-of-facial-
recognition/
Jupyter.org, 2023. JupyterLab. [Online]
Available at: https://jupyter.org/
Omkar Sawant, Y. j., 2019. researchgate. [Online]
Available at: https://ijarcce.com/wp-content/uploads/2019/05/IJARCCE.2019.8456.pdf
PranavK.B, M., 2020. researchgate. [Online]
Available at:
https://www.researchgate.net/publication/341903551_Design_and_Evaluation_of_a_Re
al-Time_Face_Recognition_System_using_Convolutional_Neural_Networks
T.Sharanya, 2020. Researchgate. [Online]
Available at:
https://www.researchgate.net/publication/342176980_Online_Attendance_using_Facial
_Recognition/link/5ee76c0d92851ce9e7e3d13d/download?_tp=eyJjb250ZXh0Ijp7ImZp
cnN0UGFnZSI6InB1YmxpY2F0aW9uIiwicGFnZSI6InB1YmxpY2F0aW9uIn19

25
22015633 Aarohan Subedi