See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/379985898

Facial Recognition-Based Attendance System

Research Proposal · April 2024

DOI: 10.13140/RG.2.2.20991.21921

CITATION READS

1 2,288

5 authors, including:

Mohammed Hamza Siddiqui

Vasantdada Patil Pratishthan College of Engineering
2 PUBLICATIONS 2 CITATIONS

SEE PROFILE

All content following this page was uploaded by Mohammed Hamza Siddiqui on 21 April 2024.

The user has requested enhancement of the downloaded file.

 Facial Recognition-Based Attendance System
Prof. Priyanka Manke, Pawan Sakat,
Vasantdada Patil Pratishthan’s College of Engineering and V.A, Vasantdada Patil Pratishthan’s College of Engineering and V.A
Dept of Information Technology, Mumbai, India Dept of Information Technology, Mumbai, India
priyankamanke@pvppcoe.ac.in, sakatpavan58@gmail.com.

Qureshi Abdul Qadir,

Mohammed Hamza Siddiqui, Vasantdada Patil Pratishthan’s College of Engineering and V.A
Vasantdada Patil Pratishthan’s College of Engineering and V.A Dept of Information Technology, Mumbai, India
Dept of Information Technology, Mumbai, India qureshiabdulqadir7869@gmail.com
hamzahsiddiqui123@gmail.com,

Himanshu Pednekar,
Vasantdada Patil Pratishthan’s College of Engineering and V.A
Dept of Information Technology, Mumbai, India
himanshupednekar100@gmail.com,

Abstract-:

Facial recognition stands as one of the most measure by identifying individuals in images and
efficient applications in image processing, playing videos. Real-world applications encompass assisting
a crucial role in the technical sphere. Identifying the visually impaired, forensic investigations,
human faces is a pressing concern, particularly in locating missing individuals, and enhancing
verifying student attendance. Utilizing facial smartphone security. This project centers on
biostatistics, an attendance system employing face developing a facial recognition-based attendance
recognition relies on high-resolution monitoring
system.. It surpasses traditional methods (location-
and advanced computer technologies. The
based, RFID, fingerprint) by reducing proxy
objective of developing this system is to digitize the
traditional method of attendance-taking, which attendance and offering validated data. The system
involves verbal calls and manual record-keeping. captures video in class, automatically matches faces
Current attendance procedures are laborious and to stored photos, and updates attendance,
time-consuming, prone to manipulation through minimizing manual work. While traditional methods
manual recording. Both traditional attendance struggle with factors like lighting and pose, the
marking and existing biometric systems are proposed system employs feature extraction and
susceptible to fraudulent proxies. This paper aims analysis for improved accuracy. It is tested under
to address these challenges. The proposed system diverse scenarios (lighting, expressions, partial
incorporates the Haar cascade algorithm, OpenCV, faces, facial variations) with recognition rate as the
Dlib, Pandas, and MySQL. Following facial key metric.
recognition, attendance reports are generated and
saved in Excel format. The system undergoes Ⅱ. LITERATURE SURVEY
testing under different conditions, such as
variations in illumination, head movements, and Our work draws inspiration from prior studies on
changes in camera-to-student distance. Rigorous facial recognition, specifically In this system, we
testing evaluates overall complexity and accuracy. adopt an approach that emphasizes facial
The proposed system proves to be an efficient and recognition efficiency, focusing on succinct two-
robust solution for classroom attendance dimensional facial features rather than intricate 3D
management, eliminating manual labour and time geometry [1]. Face recognition stands at the heart of
consumption. Additionally, the system's the recognition process, utilizing computer vision
development is cost-effective and requires minimal
technology to analyze facial features for identity
installation.
verification. Broadly speaking, it encompasses two
Keywords – OpenCV, Haar Cascade Algorithm, MySQL, primary components: face detection and face
Python. recognition matching. This technology operates by
scrutinizing the facial characteristics of individuals,
Ⅰ. INTRODUCTION processing input from face images or video streams.
Facial recognition, alongside voice, fingerprint, and [2]. During the system design phase, user
iris recognition, offers a robust biometric security requirements underwent a transformation into a
format conducive to aiding system developers in
implementation. The system design began with class is recorded if they are detected in a minimum
defining its logical structure before moving on to of 'n' snapshots, a threshold determined by the
how it would function in the real world. To achieve professor. This adaptability permits students to leave
this, we used Object-Oriented Design (OOD) during emergencies without losing attendance credit
because it offers a variety of visual aids like [7]. In contrast to the typical method, which divides
diagrams to model both the processes and the data facial features like eyes and mouth, our facial
involved. Communication resources like these are recognition system examines the entire facial area as
highly effective and easily understandable among a unified input for comprehensive analysis. This
stakeholders. [3]. Each student in the class must approach enhances our understanding of facial
register by providing necessary information. After detection in literacy assessments, unveiling a two-
registration, their images will be taken and stored in step process: initially classifying image content as
a dataset. During each class session, faces will be either facial or non-facial, then conducting a
detected from the live video stream of the classroom. thorough examination for functional interpretation
These detected faces will then be compared with and appearance analysis.
images in the dataset. If a match is found, attendance
will be marked for the respective student. At the end Ⅲ. METHODOLOGY
of each session, a list of absent students will be The face recognition process can be segmented into
emailed to the instructor. [4]. The proposed method two primary phases: pre-detection processing,
starts with enrolling students into the system. Then, encompassing facial recognition processes, and
it involves several key stages: capturing images, detection and recognition, occurring subsequently,
preprocessing them, using the Haar Cascade which involve activities such as feature extraction
classifier for face detection, creating an image and matching.
dataset, and performing face recognition using the
LBPH algorithm [5]. The architectural depiction of A. Facial localization:
the smart attendance automation system utilizing
The main goal of this stage is to identify the presence
facial recognition comprises several vital elements
and position of human faces within a provided
working together harmoniously. Commencing with
image. The results of this stage include segments
an input layer, it acquires facial images, which
containing each recognized face in the input image.
subsequently undergo processing and readiness in a
This procedure seeks to establish a facial recognition
pre-processing layer. This preliminary phase might
system that is both more reliable and simpler to
involve activities like resizing images, refining
construct.
lighting and colors, and extracting facial attributes
for recognition objectives. Once primed, the images B. Facial segmentation:
are transmitted to the facial recognition layer for
scrutiny. This layer is tasked with precisely Following face detection, the subsequent stage
recognizing faces and frequently integrates deep involves face segmentation (extraction), during
learning algorithms trained on a face dataset to which the system separates and retrieves the facial
ensure dependable identification of individuals. area from the remainder of the image. This
Fine-tuning of these models and testing for procedure entails cropping the identified faces
functionality may be included in this layer. When a according to the bounding boxes produced during
face is successfully detected, the attendance face detection. The extracted facial regions are
management layer assumes control, overseeing the subsequently standardized to ensure uniform size,
maintenance of attendance records. Activities within orientation, and lighting conditions, thereby
this layer might include recording attendance, enhancing the accuracy of subsequent processing
keeping records current, and producing reports [6]. stages.
This system employs facial recognition to
C. Face Recognition:
automatically manage student attendance.
Regarding the facial recognition algorithm as a Once facial images have been depicted, the
black box, the system determines attendance in the identification procedure commences. Establishing a
following manner: Recording initiates at the face database is crucial for enabling automated
commencement of the class, with the database recognition. Multiple photographs of each
providing the start time and room number. Every 10 individual are employed to extract their features,
minutes, a snapshot of the class is taken. Utilizing which are subsequently stored in the database.
these snapshots, the facial recognition algorithm Subsequently, the face detection and feature
identifies students and records their presence within extraction processes are carried out on an input
that 10-minute interval. A student's attendance for a
image, and the features of each facial category are 1. Eigen Faces:
compared and stored in the database.
This approach utilizes statistical methods to extract
Ⅳ. PROPOSED SYSTEM features influencing photographs. The training
database provided is essential in the entire
Efficient Attendance management is crucial for recognition process, treating images from two
ensuring operational efficiency in today’s different classes as unified entities, rather than
organisational settings. Traditional Techniques often segregated.
rely on manual processes which are susceptible to
errors and consume valuable resources. This system
proposes an innovative solution, an attendance
management system powered by facial recognition
technology, a form of biometric identification that
has emerged as a powerful tool for various
applications, including attendance tracking By
analysing your unique facial characteristics such as
key facial patches and Skin textures, facial
recognition systems can accurately identify and
verify individuals. The proposed system presents
notable benefits compared to conventional methods,
such as improved accuracy, speed, and user
convenience. The objective of the attendance
management system is to capitalize on facial
recognition technology's advantages to simplify
attendance tracking in organizations and institutions. 2. Linear Discriminant Analysis (LDA).:
By integrating this technology organisations can
achieve higher levels of accuracy and efficiency in Like Eigenfaces, the Linear Discriminant Analysis
keeping track of employee or student attendance (LDA) Also known as Fisher faces algorithm
Additionally, the contactless nature of the facial follows an iterative approach, incorporating
recognition eliminates the need for physical principal components analysis as a variation of
identifiers like id cards, offering enhanced security Eigenfaces. The key difference lies in how the
and user convenience. The proposed system is algorithm handles classes. Unlike Eigenfaces, which
developed using Haar classifiers the machine doesn't differentiate between images from separate
learning approach that will be employed for facial classes during training, Fisher Faces employ linear
detection within images. Local binary pattern discriminant analysis to distinguish between images
histograms (LBPH) will further enhance the facial of distinct classes. While each image is influenced
recognition by capturing local texture patterns in by the overall average, Fisher Faces enable
facial images. Furthermore Python a versatile discrimination among images of different classes.
programming language will be used due to
simplicity, readability and extensive libraries for
machine learning and image processing tasks.
Finally, structured query language(SQL) will be
used to show facial images and the associated data
in the database, enabling efficient retrieval and
management of attendance records.
D. Types of Algorithms:

This system employs various types of algorithms,

including:
3. Local Binary Feature Histograms:
1.Eigen Faces.
Local Binary Feature Histograms serves as a crucial
2.Linear Discriminant Analysis (LDA). texture descriptor employed in image analysis. Its
3.Local binary pattern histograms. functionality involves a comparison of individual
pixels with their adjacent counterparts. Essentially,
4. Haar Classifiers. LBP aims to depict the inherent local structure of an
image by encoding the interplay between the central
pixel's intensity and that of its neighbouring pixels.
This encoding process commonly entails the
application of thresholding to pixel values, thereby
generating binary patterns. Afterward, these binary
patterns undergo a transformation into decimal
values, enabling the generation of a histogram
representation. Within the domain of facial
recognition attendance management systems, the
importance of effective feature extraction methods
becomes paramount in ensuring accurate
identification of individuals from facial images. Ⅴ. Block Diagram

4. Haar Classifiers:
Ⅵ. FLOWCHART
The Haar classifier, commonly referred to as a Haar
cascade classifier, is an object detection algorithm
based on machine learning, utilized for identifying
objects or patterns within digital images. Its most
prominent use is in face detection, although it can
also be trained to recognize a range of other objects.
Haar features operate akin to convolutional kernels
and are instrumental in detecting features within an
image. These features encompass various types,
including line features, edge features, and four-
rectangle features, among others. Each feature is
quantified by a singular value derived from the
contrast between the sums of pixels beneath
corresponding black and white rectangles. Haar
cascades offer several advantages. Their reliance on
simple Haar features and the cascading structure
make them computationally efficient, enabling real-
time object detection in images and video streams.
Additionally, the algorithm can detect objects of
different sizes by resizing the search window.
Furthermore, Haar cascades can be trained to detect
a wide variety of objects by using appropriate
training datasets, making them adaptable.
The training process begins by examining the
directory containing the training data. Each image in
the training dataset is initially converted to B. PANDAS
grayscale. Next, the center of each image is
pinpointed, and a comparison against its Pandas, a Python library available as open-source,
surrounding regions is conducted using a provides a wide range of tools for conducting data
predetermined threshold. If the intensity of the analysis. Within the package, numerous data
center portion equals or exceeds that of its structures are available, suitable for a broad
neighbouring areas, it is designated as 1; otherwise, spectrum of data manipulation tasks. Additionally, it
it is labelled as 0. Following that, the subsequent encompasses diverse data analysis algorithms,
enabling the resolution of data science and machine
steps include resizing the images and converting
learning tasks using Python.
them into a numpy array, which acts as the
fundamental data structure of the numpy library.
C. Visual Studio Code
Following this, each face within the photograph is
identified and sorted into distinct lists, each
accompanied by its corresponding ID. Subsequently, Visual Studio Code (VS Code) stands out as a
the faces are trained alongside their respective IDs. versatile Integrated Development Environment
To clarify this process, another flowchart depicting (IDE) known for its adaptability and effectiveness.
the system's operation is necessary. It is created by Microsoft., it offers a smooth
experience across various platforms, including
The system functions in the following manner: upon Windows, macOS, and Linux. Utilizing TypeScript
reading the image, it is converted into grayscale. The and built on the Electron framework, VS Code
Haar Cascade frontal face module is employed to provides a reliable foundation for developers with
detect faces within the image. Prediction of the faces diverse skills and programming backgrounds. At its
in the image is performed using the LBPH core, VS Code features a highly customizable multi-
window text editor designed to enhance
algorithm. Once the faces are predicted, they are
productivity. Equipped with support for multiple
outlined within a green box along with their
undo actions, syntax highlighting for numerous
corresponding names. programming languages, and intelligent indentation,
Ⅶ. SOFTWARE USED coding becomes a seamless and intuitive process.

A. Open CV
D. Microsoft Excel
Upon being captured, the input image undergoes
processing by the phone's camera, immediately
converting it to grayscale. Utilizing the Haar The Microsoft Office suite comprises Microsoft
Cascade frontal face module, the system proceeds to Excel, a spreadsheet tool. Spreadsheets offer users the
detect faces within the image. Subsequently, the ability to arrange data into tabular formats, organized
LBPH algorithm is applied to predict the detected in rows and columns, facilitating mathematical
faces. Once predicted, identified faces are outlined manipulation through both fundamental and
in a green box alongside their respective names. sophisticated arithmetic operations. In addition to its
OpenCV (Open Source Computer Vision Library) is conventional spreadsheet functionalities, Microsoft
an open-source software library primarily focused Excel features powerful graphing and charting tools,
on machine learning for computer vision tasks. It supports programming through Visual Basic for
provides support for various platforms including Applications (VBA), and enables data retrieval from
Windows, Linux, Android, and macOS, and offers external sources using Dynamic Data Exchange
interfaces in languages such as C++, Python, Java, (DDE). Excel, tailored for the creation of electronic
and MATLAB. Real-time vision applications are a spreadsheets, serves as a platform for managing,
significant aspect of OpenCV's capabilities, making editing, and storing data efficiently. Early electronic
use of SSE and MMX instructions whenever spreadsheet systems drew inspiration from paper
possible. Ongoing development efforts include the spreadsheets commonly utilized for accounting
creation of fully functional CUDA and OpenCL purposes, thus retaining a similar foundational design.
interfaces. With a wide array of algorithms and Computerized spreadsheets mirror the basic structure
supporting functions, OpenCV is predominantly of paper spreadsheets, with related data organized into
implemented in C++ and features a template tables consisting of standardized rows and columns of
interface seamlessly integrated with STL containers. small rectangular boxes or cells
 Ⅷ. FUTURE SCOPE

In summary, the integration of AI and machine

In the foreseeable future, the integration of AI and learning into the attendance management system
machine learning into this attendance management promises heightened accuracy, efficiency, and
system presents a significant opportunity for security. This evolution represents a mutually
advancement. Envision a facial recognition system beneficial advancement, facilitating smoother
that achieves heightened levels of accuracy and attendance tracking and streamlined operations for
reliability. Through the incorporation of all stakeholders involved.
sophisticated AI algorithms such as deep learning,
the system would adeptly identify faces even in Ⅸ. RESULTS
challenging scenarios, such as low lighting or A. Main Screen
unconventional angles. This refinement would result
in diminished errors and heightened user
confidence. Furthermore, machine learning
capabilities could enable the system to discern
attendance patterns over time. It would have the
capacity to recognize anomalies, such as consistent
tardiness or unexpected absences, thereby
facilitating more effective attendance management
and ensuring operational efficiency. Moreover, the
potential for predictive analytics is substantial.
Envisage the \system's ability to forecast attendance
trends based on historical data. This predictive
capability could revolutionize scheduling and
resource allocation strategies, allowing for B. Taking Attendance
optimized operations and heightened productivity.

The integration of IoT devices offers another

dimension of automation. Consider the possibility of
sensors or smart locks autonomously detecting
individuals' arrivals and departures, obviating the
need for manual check-ins and streamlining
attendance tracking processes. Additionally, the
system's continuous learning capabilities promise
ongoing improvement. Through iterative refinement
and adaptation to evolving attendance patterns and
environmental factors, its effectiveness would
endure over time, perpetually enhancing and fine-
tuning its algorithms.

Security enhancements are also within reach. Facial

recognition technology could extend beyond
attendance management to fortify access control
measures. Imagine restricted areas being accessible
only to authorized individuals, thereby bolstering C. Attendance Taken
organizational security protocols. On the
administrative front, integrating the system with HR
and payroll functions offers substantial benefits.
Automated data transfers between systems would
mitigate errors and streamline processes related to
payroll, leave management, and compliance.
 D. Recorded Attendance [6] M. Shamila1, Bhanu Prakash, Asrar Ahmed,
Poshak Prajeet, Ruby Pant ”Smart Attendance
Autiomation System” Department of CSE (AIML),
GRIET, Hyderabad, India Uttaranchal Institute of
Technology, Uttaranchal University, Dehradun,
248007, India.

[7] Ashwin Rao “AttenFace: A Real Time

Attendance System Using Face Recognition”
International Institute of Information Technology,
Hyderabad ashwin.rao@students.iiit.ac.in

[8] Dr. Sarita Sanap, Ms. Sakshi Narwade, Mr. Sahil

Goge, Mr. Krishna Pandit “Face Recognition Based
Ⅹ. CONCLUSION Attendance System Using Histogram of Oriented
The Attendance Management System, powered by Gradients and Linear Support Vector Machine”
facial recognition, provides students with quick and Assistant Professor, Department of ETC
convenient access to their attendance data, which is Engineering, MIT, Aurangabad, Maharashtra, India.
subsequently transferred to an Excel spreadsheet. [9] S. Dev and T. Patnaik, "Student Attendance
the proposed system offers improved facial System using Face Recognition," 2020 International
recognition process that is not dependent on any Conference on Smart Electronics and
external devices like raspberry pi or any other Communication (ICOSEC), Trichy, India, 2020, pp.
module for capturing faces and recording attendance 90-96.
[1]. This system offers technology with improved
security and user friendly interface. [10] Mihir Ghanekar, Archita Sehgal, Shrishail
Gouraond, Maitray Wani, Prof. Pramila M “Facial
Ⅺ. REFERENCES recognition based attendance system” International
[1] A. Arjun Raj, M. Shoheb, K. Arvind and K. S. Research Journal of Engineering and Technology,
Chethan, "Face Recognition Based Smart Dept of Computer Engineering and IT, Mumbai,
Attendance System," 2020 International Conference Maharashtra, India.
on Intelligent Engineering and Management [11] Salim, Omar Abdul Rhman, Rashidah Funke
(ICIEM), London, UK, 2020, pp. Olanrewaju, and Wasiu Adebayo Balogun. "Class
[2] H. Yang and X. Han, "Face Recognition attendance management system using face
Attendance System Based on Real-Time Video recognition." 2018 7th International Conference on
Processing," in IEEE Access, vol. 8, pp. 159143- Computer and Communication Engineering
159150, 2020. (ICCCE). IEEE, 2018.

[3] Edison Kagona and Sani Usman "Student’s [12] Siswanto, Adrian Rhesa Septian, Anto Satriyo
Attendance Management in higher institutions using Nugroho, and Maulahikmah Galinium.
azure cognitive service ad Open CV face detection "Implementation of face recognition algorithm for
& recognition attendance system” Department of biometrics based time attendance system." 2014
Computer Science & Information Technology, International Conference on ICT For Smart Society
Faculty of Science & Technology, International (ICISS). IEEE, 2014.
University of East Africa; P.O.Box 35502 Kampala. [13] Sen, S., Patel, M., Sharma, A.K. (2021).
[4] Smitha, & Hegde, Pavithra & Afshin,. (2020). Software Development Life Cycle Performance
Face Recognition based Attendance Management Analysis. In: Mathur, R., Gupta, C.P., Katewa, V.,
System. International Journal of Engineering Jat, D.S., Yadav, N. (eds) Emerging Trends in Data
Research and. V9. 10.17577/IJERTV9IS050861. Driven Computing and Communications. Studies in
Autonomic, Data-driven and Industrial Computing.
[5] Sandhya Potadar , Riya Fale , Prajakta Springer, Singapore.
Kothawade , Arati Padale, 2021, Attendance
Management system using Face Recognition,
INTERNATIONAL JOURNAL OF
ENGINEERING RESEARCH & TECHNOLOGY
(IJERT) Volume 10, Issue 08 (August 2021).

View publication stats