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

net/publication/380296129

Facial Recognition Based Attendance System Using Opencv

Article · May 2024

CITATION READS
1 1,798

6 authors, including:

Vyshnavi Ms Tjcse Dheemansh Publication Hub

Heriot-Watt University The Journal of Computational Science and Engineering
1 PUBLICATION 1 CITATION 31 PUBLICATIONS 14 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Tjcse Dheemansh Publication Hub on 03 May 2024.

The user has requested enhancement of the downloaded file.

 The Journal of Computational Science and Engineering.
ISSN: 2583-9055
Facial Recognition Based Attendance System
Using Opencv
Ms. K. Vyshnavi 1, Suhel Shaik 2, Mylavarapu Santosh Kumar 3,
M.S.V Praveen4, Swetha Potti5
1,2,3,4,5
Department of Computer Science and Engineering,
Nadimpalli Satyanarayana Raju Institute of Technology, Visakhapatnam-531173,
vyshnavik.cse@nsrit.edu.in1, shaiksuhel2000@gmail.com2,
santoshkumarmylavarapu@gmail.com3, msvpraveen2@gmail.com4,
swethapotti2001@gmail.com5

Keyword: ABSTRACT
Face Recognition,
Taking attendance is one of the critical tasks that need to be completed every
OpenCV, day in businesses, schools, colleges, and organizations. Most of the time, it is
LBPH, done by hand using methods like calling someone by name or roll number. The
main aim of this project is to design an automated attendance system based on
Haar cascade. face recognition technology, intended to supersede the current manual
procedure. This project satisfies the time management requirements as well as
the requirements for updating the attendance system. This device is placed in
the classroom and is used to train students' data, including name, roll number,
class, sec, and photo. An OpenCV is used to extract the photos. The student can
approach the machine before the commencement of the relevant class, and it
will start taking photographs and comparing them to the eligible dataset. The
device automatically records attendance by identifying faces using both the
Haar cascade algorithm and the LBPH (Local Binary Pattern Histogram)
Algorithm. It then compares the histogram data to a pre-existing dataset. An
Excel spreadsheet is created and updated hourly using data from the relevant
class instructor.

INTRODUCTION

The previous attendance management system suffered from significant shortcomings,

primarily revolving around data accuracy, time efficiency, and accessibility. Firstly, the
system's reliance on manual recording opened avenues for inaccuracies, as attendance could
be marked by proxies without institutional awareness, compromising data integrity. For
instance, a scenario where one student signs in for another undermines the system's reliability
for analytical purposes. Moreover, enforcing stricter measures to address this would entail
considerable human resources and time, rendering it impractical.

Secondly, the system proved excessively time-consuming, with each student requiring
approximately a minute to sign their attendance on multi-page lists. This inefficiency resulted
in only around 60 students being processed per hour, severely hampering productivity.

Lastly, the lack of accessibility for concerned parties, such as parents eager to monitor
their child's attendance, further highlighted the system's limitations. Parents lacked means to
The Journal of Computational Science and Engineering. ISSN: 2583-9055

Volume: 2 Issue: 3 May 2024 Page : 1

 The Journal of Computational Science and Engineering.
ISSN: 2583-9055
access such crucial information, underscoring the need for system evolution to enhance
efficiency, data accuracy, and accessibility for legitimate stakeholders.

Today, many institutions use automated systems for creating attendance records. RFID
technology and biometric techniques are two examples of such systems. Despite being
automatic and more advanced than the conventional method, it is unable to meet the time
constraint. The pupil must stand in queue to receive attendance, which is time-consuming. The
approach can also be used in other teaching or exam contexts. activities for which presence is
necessary. With this system, traditional student identity, such as calling the student's name or
confirming the student's identification cards, may not only be the continuous process of
instruction, but it can also cause anxiety in students during examination times. Furthermore,
pupils must register in the database. to be acknowledged. The user-friendly interface allows
for immediate enrollment.

LITERATURE SURVEY

To create an efficient task system, researchers have tested and analyzed various
factors related to camera-based total item detection, identification, and distance estimates. This
study particularly focused on the photo identification phase and analyzed it thoroughly.
The majority of the research articles have concluded that CNN is the most widely used
technology for photo identification. Although some tactics were found to be effective, others
had obstacles. The authors of this paper suggested a way to integrate facial recognition
technology with MATLAB GUI and the basic component analysis (PCA) method, utilizing the
Eigenface database. This will automate the attendance system. The device initiates by
capturing a student's picture, preprocessing it, and then applying the Eigenface database.
Subsequently, it compares the captured face image with an Eigenface image. Attendance is
recorded when the similarity distance score falls below the threshold value of 0.3. The results
are saved in a Microsoft Excel sheet that is integrated with the MATLAB GUI. The
preliminary face database has ten pictures of each of the fifteen individuals with varying
positions and instructions.

The authors of this research have used face verification and radio frequency
identification (RFID) to design and execute a classroom attendance system. The students are
identified through their RFID cards, and for further identity confirmation, a face recognition
method called the Rapid Adaptive Neural Network Classifier (FANNC) has been included.
The classifier's ability to recognize human faces has been tested and taught. Each student
needs to take seven pictures with different head postures for their images to be recognized by
the classifier. The facial recognition system was tested on six different student images, and for
the front face, it was accurate up to 98% of the time[3].

The authors aimed to modernize the outdated manual attendance system with a
digital alternative integrating facial recognition technology. MATLAB software was utilized to
deploy the principal component analysis (PCA) algorithm for the face recognition module.
Upon successful facial authentication, the code was transferred to an embedded hardware
device via a Microcontroller PC, which was also linked to a servo motor for unlocking the
door. Testing indicated the system's heightened sensitivity to variations in background and
The Journal of Computational Science and Engineering. ISSN: 2583-9055

Volume: 2 Issue: 3 May 2024 Page : 2

 The Journal of Computational Science and Engineering.
ISSN: 2583-9055
head orientation[4].

OBJECTIVE:

To precisely recognize the faces of the students. to automatically record attendance. To

cut down on the time and labor needed for manual attendance to give teachers and students
access to a useful system of attention. It cuts down on time loss and offers flexibility. A proxy
will not be an option
The following results are anticipated to meet the goals
1. To identify the facial segment from the video frame.
2. To take the identified face and extract its useful traits.
3. To categorize the characteristics to identify the discovered face.
4. To document the designated student's attendance.

Fig-1: Block diagram

EXISTING SYSTEM

Over the past two decades, computer-based facial recognition has witnessed significant
advancements. Notably, various algorithms have been developed to a degree where computers
can rival human accuracy in facial recognition. To develop our solution, it's imperative to
analyze the mechanisms behind facial recognition, evaluate existing algorithms, and explore
the diverse applications of this technology.

Regrettably, Sinha, Strovsky, and Ussell have presented nineteen key insights into
human facial recognition, highlighting numerous techniques employed by individuals in
identifying faces. Their findings underscore the intimate connection between the study of
human facial recognition processes and the artificial algorithms utilized in facial recognition
systems. Remarkably, facial recognition in the human brain occurs in just 1 millisecond.

The Journal of Computational Science and Engineering. ISSN: 2583-9055

Volume: 2 Issue: 3 May 2024 Page : 3

 The Journal of Computational Science and Engineering.
ISSN: 2583-9055

PROPOSED WORK AND ALGORITHMS

Fig-2: System Architecture

Every student in the class must register by providing necessary information, followed
by capturing their photos to include in the dataset. The system will then identify faces from the
captured images and compare them to pictures in the dataset. Upon finding a match, the
system will record the student’s attendance. The proposed system is tasked with photographing
each student’s face and storing it in the attendance database. It’s crucial that the photograph
captures every facial feature, including the student’s posture and sitting position

Algorithms
1. Haar cascade Algorithm:
In this machine learning approach, a cascade function undergoes training using a
substantial dataset comprising both positive and negative images. Positive images depict
instances where the target object is present, while negative images showcase its absence.
Subsequently, this trained cascade function is utilized to detect objects in additional images.
Thankfully, OpenCV offers pre-trained Haar cascade algorithms, conveniently categorized
(e.g., faces, eyes, etc.) based on the images utilized during training.

The Journal of Computational Science and Engineering. ISSN: 2583-9055

Volume: 2 Issue: 3 May 2024 Page : 4

 The Journal of Computational Science and Engineering.
ISSN: 2583-9055

Fig-3: Haar Features

LBPH Algorithm:

The Local Binary Pattern Histogram (LBPH) texture operator is a simple yet
remarkably powerful method. It assigns a binary value to each pixel in an image based on
comparing its intensity with the surrounding pixels. Additionally, in certain datasets,
combining LBP with the histograms of oriented gradients (HOG) descriptor has shown a
notable enhancement in detection accuracy.

Fig-4: LBPH Algorithm

IMPLEMENTATION STUDY

Modules

1. OpenCV:

OpenCV stands as an expansive open-source toolkit catering to image processing,

machine learning, and computer vision tasks. Compatible with various programming
languages like Python, C++, Java, and more, OpenCV boasts versatility. Its capabilities span
from recognizing faces and objects to deciphering human handwriting through the analysis of
images and videos. The integration of additional libraries, such as NumPy, amplifies its
functionality. OpenCV seamlessly incorporates the diverse range of operations achievable with
The Journal of Computational Science and Engineering. ISSN: 2583-9055

Volume: 2 Issue: 3 May 2024 Page : 5

 The Journal of Computational Science and Engineering.
ISSN: 2583-9055
NumPy, further enriching its array of tools and functionalities.

With the aid of a vast array of OpenCV projects and applications, this course will
assist you in learning the fundamentals to the advanced levels of image processing, including
operations on images and videos.

1. Tkinter:
The Tk/Tk GUI toolkit's standard Python interface is provided by the tkinter package,
sometimes known as the "Tk interface." Windows computers and the majority of Unix
platforms, including macOS, support Tk and tkinter. You can verify that tkinter is correctly
installed on your system by using the command line command python -m tkinter. This will
create a window that shows a basic Tk interface.

2. PIL Library:
An older library known as PIL was forked to create the Python Pillow library. The Python
Imaging Library, or PIL for short, was the first library that allowed Python to work with images. PIL
only supports Python 2 and was deprecated in 2011. Pillow, as described by its creators, is a PIL fork
that is friendly and supports Python 3, helping to keep the library alive. Python has multiple modules
for handling images and carrying out image processing. You can use NumPy and SciPy to work
directly with images by modifying their pixels. Other well-liked image processing libraries include
Mahotas, scikit-image, and OpenCV. Compared to Pillow, several of these libraries are quicker and
more potent.
3. Time Module:
To manage tasks involving time, Python has a module called time. We must import the module
before we can use the functions it defines.

4. Date Time Module:

While Python does not inherently possess a distinct data type for dates, we can effectively
manipulate dates by importing the datetime package and utilizing date objects.
5. NumPy:
"NumPy" is short for "Numerical Python," and it is a Python package that serves as the
fundamental library for scientific computing. It includes tools for integrating other languages such as C
and C++. NumPy also provides a robust n-dimensional array object, and it can help with tasks like
random number generation and linear algebra. The NumPy array is another useful multi-dimensional
container for storing generic data. To install NumPy for Python, type "pip install numpy" at the
command prompt, and once the installation is complete, open your integrated development
environment (IDE), such as PyCharm, and import NumPy as "np.".
6. Pandas:
Using its robust data structures, Pandas is an open-source Python library that offers
high-performance data manipulation and analysis capabilities. Panel data is the word that refers to an
econometric analysis of multidimensional data, hence the name Pandas.Panda's module is not included
in the standard Python distribution.

The Journal of Computational Science and Engineering. ISSN: 2583-9055

Volume: 2 Issue: 3 May 2024 Page : 6

 The Journal of Computational Science and Engineering.
ISSN: 2583-9055
METHODOLOGY
The student attendance system implementation revolves around face recognition. It
initiates with image capture via a user-friendly interface, followed by preprocessing of
captured facial images. Features are then extracted from these images, leading to subjective
selection and classification for recognition. Extensive evaluation and computation are
conducted on both Local Binary Pattern (LBP) and Principal Component Analysis (PCA)
feature extraction methods for comparison purposes. Notably, LBP is enhanced in this
approach to counter illumination effects. Additionally, an algorithm is developed to combine
enhanced LBP with PCA, focusing on subjective selection to bolster accuracy. Further details
on each stage will be illustrated in the subsequent diagram.

Fig-5: Flow-chart of the methodology

Face detection relies on the Haar Cascade algorithm, where a cascade function is
trained using a multitude of positive and negative images. In this context, images containing
faces are labeled as positive, while those without faces are labeled as negative. The features
utilized for face detection are essentially data extracted from photographs, enabling
differentiation between various images. During training, all images undergo evaluation
against each algorithmic feature, with equal initial weighting assigned to each image. The
optimal threshold for classifying faces as positive or negative is determined, although
misclassifications and errors may occur. Features with the lowest error rates are selected,
indicating their effectiveness in distinguishing between images with and without faces.

The Journal of Computational Science and Engineering. ISSN: 2583-9055

Volume: 2 Issue: 3 May 2024 Page : 7

 The Journal of Computational Science and Engineering.
ISSN: 2583-9055
RESULTS

Fig-7: Project Interface

Fig- 6: Attendance before updating

Fig-9: Predicted Result

Fig-8: Registering the candidate

Fig-10: Attendance after updated

The Journal of Computational Science and Engineering. ISSN: 2583-9055

Volume: 2 Issue: 3 May 2024 Page : 8

 The Journal of Computational Science and Engineering.
ISSN: 2583-9055
Fig-11: Attendance marked in Excel sheet

CONCLUSION:

The Automated Attendance System aims to rectify the errors inherent in traditional
manual attendance methods by automating the process. This system is designed to be
adaptable for use in various institutional settings, offering a correct and efficient alternative to
outdated manual techniques, particularly in office environments. It ensures reliability, safety,
and practicality without the need for specialized hardware, as it can be implemented using
standard computer and camera equipment. By facilitating attendance tracking in lectures,
sections, or laboratories, it streamlines the process for lecturers or teaching assistants,
particularly beneficial in managing large class sizes. The Automated Attendance method
addresses the limitations of manual methods, showcasing the application of image-processing
techniques in educational settings. Its implementation not only improves efficiency but also
enhances an institution's reputation.

FUTURE SCOPE

Future developments in facial recognition-based attendance systems with OpenCV

have a great deal of promise. Further research and development can concentrate on improving
accuracy using deep learning techniques, which will allow the system to consistently identify
people in a variety of circumstances, including changing lighting and facial expressions. Faster
attendance tracking and analysis will be possible because of real-time processing optimization,
which will allow the system to manage massive volumes of data efficiently. The system's
capabilities will be further enhanced by integration with artificial intelligence, which will
enable it to adapt to and learn from fresh data, gradually increasing accuracy. It will be
imperative to address privacy and security concerns, which calls for the use of strong
encryption techniques and adherence to legal requirements. Investigating the integration of
multiple biometric modalities, such as merging facial recognition with additional biometric
modalities.

ACKNOWLEDGMENT

We are deeply grateful to our mentor Ms. K. Vyshnavi for the guidance,
encouragement, and constructive feedback at every stage of this project. Your knowledge and
expertise greatly aided the project's growth and quality.

REFERENCES

1. "Smart Attendance System Using Face Recognition," International Journal of Research in Advanced
Computer Science Engineering (IJRACSE), Volume 7 Issue 2, July 2021, Pages 10–18, by N.
Viswanadha Reddy, K. Roshini, Prashant Mishra, G. Thulasi Tirumaleswari, and Y. Durga Sai Chandu.

The Journal of Computational Science and Engineering. ISSN: 2583-9055

Volume: 2 Issue: 3 May 2024 Page : 9

View publication stats

The Journal of Computational Science and Engineering.

ISSN: 2583-9055
2. "Automated Smart Attendance System Using Face Recognition," Preethi, Kolipaka, and Swarthy
Vodithala. In the ICICCS 5th International Conference on Intelligent Computing and Control Systems,
2021, pp. 1552-1555. IEEE, 2021.

3. O. Shoewu and O.A. Idowu, Development of Attendance Management System using Biometrics ",
The Pacific Journal of Science and Technology, Vol. 13, Number1, pp.300- 307, May 2012 (Spring).

4. Damir Demirovic, Emir Skejic, Amira Serfovic, Performance of some images processing algorithms
in TensorFlow, computer vision and pattern recognition, pp, 799- 778, 2018.

5. Raj, A. Arjun, K. Arvind, Mahammed Shoheb, and K. S. Chethan a smart attendance system that
uses face recognition. In ICIEM 2020: International Conference on Intelligent Engineering and
Management, pp. 354–357. IEEE, 2020.

6. "Real time automatic attendance system for face identification using face API and OpenCV," by
Sikandar Khan, Adeel Akram, and Nighat Usman. 2020: 469–480 in Wireless Personal
Communications 113, no. 1.

7. Hl, Neha Kumari Dubey, K. Vishal, M. R. Pooja, and Dhanush Gowda. "Attendance system based
on face recognition." IJERT 9, no. 6: International Journal of Engineering Research & Technology
(2020).

8. Michael Dobson, Douglas Ahlers, Bernie DiDario, 2018 Fourth International Conference on
Computing Communication Control and Automation (ICCUBEA). IEEE, 2018.

9. Mothwa, Louis, Jules-Raymond Tapamo, and Temitope Mapati. "Conceptual model of the smart
attendance monitoring system using computer vision." In 2018 14th International Conference on
Signal-Image Technology & Internet-Based Systems (SITIS), pp. 229-234. IEEE, 2018

10. Bussa, Sudhir, Ananya Mani, Sakshi Kaushik, and Shruti Bharuka. "Facial recognition-based smart
attendance system utilizing OPENCV." International Journal of Engineering Research and Technology,
Volume 9, Number 3, 2020

11. http://www.openCV.org [6] Hapani, Smit, et al. "Automated Attendance System Using Image
Processing." 2018 Fourth International Conference on Computing Communication Control and
Automation (ICCUBEA). IEEE, 2018.

The Journal of Computational Science and Engineering. ISSN: 2583-9055

Volume: 2 Issue: 3 May 2024 Page : 10