Face Recognition Attendance System

Dr. Ravindra Nath Suryanarayana

Babasaheb Bhimrao Ambedkar Babasaheb Bhimrao Ambedkar
University,Lucknow, Uttar Pradesh University, Lucknow, Uttar Pradesh
India India
rnkatiyar@gmail.com suryanarayanbaijpur@gmail.com

Rishi Maurya Anuj Saroj

Babasaheb Bhimrao Ambedkar Babasaheb Bhimrao Ambedkar
University,Lucknow, Uttar Pradesh University,Lucknow, Uttar Pradesh
India India
rishimaury893@gmail.com sarojanuj213@gmail.com

Abstract
This research paper provides an in-depth exposition on the construction of a
face recognition attendance system. The proposed system intends to automate
the attendance task through utilizing different techniques of facial recognition to
recognize and verify registered individuals. The system uses a camera to capture
images of individuals, and the collected data is processed and analyzed with
a variety of image processing techniques to extract facial characteristics. The
extracted characteristics are then compared with the existing base to verify the
individual's identity. The system generates attendance reports notifying teachers
or managers to monitor, and manage, attendance accordingly in real-time. The
paper mentions the approach and experiments set up, the images processing tech-
niques used and the algorithms applied to reach the objective. The experimental
results indicate the feasibility and potential success of the system at respectively
identifying individual faces and generating attendances automatically. Finally,
the paper highlights the benefits of an automated face recognition attendance
system, such as time-saving, error reduction, and management possibilities for
attendance in organizations.

1
Keywords- Face Recognition, Attendance Management System, Haar Cascade,
Open CV.
1. Introduction:
Historically, the processes for measuring attendance in classrooms, meetings
and workplaces have been time-consuming, manual, and prone to errors. The
standard methods have varied, but they often been based on pen-and-paper,
roll call, or arguments regarding scanning barcode assignments that may be
inaccurate, inaccurate record-keeping may slow transitions or result in lost
or improperly recorded information. Face recognition technology has been
improving to stabilize automated attendance methods, allowing for the fastest
and most reputable attendance measurements in an unimpeachable manner.
The face recognition attendance system being proposed leverages modern com-
puter vision and machine learning methods to recognize and verify identifying
individuals through facial recognition. The proposed system uses a camera
to capture images of individuals, and the captured information is processed
through many image processing methods to recognize and then model facial
images. The facial models are used to verify the identity of each individual, and
then attendance is effectively captured in a collection. The proposed system
would effectively be available in real time and able to provide reports and allows
teachers or managers to track attendance effortlessly.
The research study outlined here would provide detailed descriptions of the
proposed face recognition attendance system and will add on to the proposed
advantages of the new system in relation to attending methods compared to the
old. Afterwards the paper will elaborate examining the methods and descriptions
the methods use in deep learning algorithms to develop this system. The research
will gather performance data and long-term usage data while exploring the
ability to monitor facial expressions while attending, and the impact of lighting
or shadows over face recognition. In the end, the study will consider other
applications of a proposed facial recognition attendance system for various
settings such as classrooms and workplaces, while considering benefits affecting
everyone.
2. Literature Survey:
Numerous investigations have been undertaken into the implementation of facial
recognition technology as a component of attendance systems. For example, a
study by Zhang et al. (2018) suggested a facial recognition attendance system
that incorporated deep learning and artificial intelligence algorithms. The study
employed a convolutional neural network (CNN) model to identify recognized
facial images, achieving a recorded accuracy of 97.5% on a dataset of 600 facial
images.
Moreover, Malyala et al. (2019) also presented a hybrid approach to attendance
management based on the utilization of RFID and facial recognition technologies.
In this system, RFID tags recorded attendance, but facial recognition verified

2
the identity of individuals, thus reducing errors and preventing fraud.
Similarly, Lee et al. (2019) implemented a facial recognition attendance sys-
tem that used a smartphone camera and cloud-based architecture. The study
concluded successfully demonstrating that the system operated with high ac-
curacy confirming abnormal attendance circumstances, and provided real-time
attendance records monitoring.
In addition, Intharawijitr et al. (2019) presented an extensive assessment of
various face recognition algorithms including Eigenface, Fisherface, and Local
Binary Patterns Histogram frame alerts (LBPH) in attendance systems indicating
that LBPH provided the most accurate results with high efficiency in a facial
recognition attendance system.
The above studies of the literature effectively highlight the considerable potential
of facial recognition technology in attendance management systems. In addition,
these studies suggest that facial recognition technologies are effective not only
at identifying individuals but also provide an effective, fast, reliable, and secure
approach to manage attendance data.
3. Software Used:
OpenCV Python Software: OpenCV (Open Source Computer Vision Library)
is a Python library for image processing and computer vision. It provides a
range of functions from object detection and facial recognition to tracking.
This makes the installation of OpenCV very important. To install OpenCV,
open your command prompt and type python. It will return your version of
Python. Next, type "pip install OpenCV-python."
VS Code: Visual Studio Code, commonly referred to as VS Code, is a free
open-source code editor created by Microsoft. It supports debugging, syntax
highlighting, intelligent code completion, snippets, code refactoring, and embed-
ded Git. A user may change its themes, keyboard shortcuts, preferences, and
install extensions that add their own functionalities.
Tkinter: Tkinter is the standard GUI library for Python. Python with Tkinter
provides an easy and fast way to create GUI applications. Tkinter is a powerful
object- oriented interface to the Tk GUI toolkit. It is a very simple task to
create a GUI application with Tkinter.
MySQL Database: Database management and creation is undoubtedly the
most important step. A database is needed to take images of every student in
a given class. The basic process of creating a database is to create a system
that takes images. For that, we will be using a camera module, and pictures of
each student registered in a particular class will be taken. The image processing
techniques used thereon will include the making of the data understandable by
a computer, all leading to saving it in a .csv file
4. Algorithm Used:

3
Haar Cascade Algorithm: The Haar Cascade Algorithm is a machine learning
object-detection algorithm for detecting objects in images or videos, mainly faces.
This consists of finding a toy classifier by employing a very large number of
positive and negative image datasets. The major steps involve:
Training: Train the algorithm over an immense dataset containing
labeled images. Positive images carry an object to be detected (like
faces), while negative ones do not.
Haar features: These are nothing more than quite simple rectangu-
lar features to represent the discontinuities in the general distribution
of pixel intensities or in edges. These are quite efficient to compute;
they are therefore capable of extracting quite simple features such as
edges and lines.
Cascade Classifier: A number of increasingly complex stages, to
quickly filter out negative windows and focus computation on promis-
ing areas. This cascade accelerates the speed for higher detections
with reasonable accuracy.
Detection: Scan the image with the trained classifier in multiple
scales and relative positions for object detection.
Local Binary Pattern Histogram (LBPH): The Local Binary Pattern His-
togram (LBPH) algorithm is classified under face recognition methods, analyzing
the texture of an image based on the neighborhood information of each pixel,
representing it in the form of binary patterns. Such patterns are used to create
Northwind histograms that represent texture, and comparisons of these his-
tograms can be done for face recognition. Owing to its simplicity and efficiency,
LBPH finds a prime place in almost all applications involving face recognition
in near real-time.
Applying the LBP operation: The process is similar to that of computing a
Local Binary Pattern (LBP) operation for texture analysis in computer vision.
For every pixel of the grayscale image, a comparison is made to the pixels
surrounding it, within a defined radius. A reading of 0 or 1 is then assigned to
each of the surrounding pixels' strength based on whether or not it is lower than
that of the impressed pixel. These bits are then concatenated to form a single
binary number that represents the central pixel. The binary number is generally
turned into a decimal value which will then represent the texture pattern of
that local area. The whole process is carried out for each pixel of the image,
producing a new image in which each pixel value corresponds to that of the local
binary pattern code. A histogram for that LBP image can then be constructed,
each bin in which half the number of pixels in the image was assigned that
local binary pattern code. The histogram can create a very compact description
of the texture contained in the image for later use in face recognition, texture
classification, or image retrieval. The LBP operation is efficient, monotonic
illumination changes resistant, and computationally efficient, making it useful
for real-time applications.

4
Extracting the Histograms: We shall now describe another related method
called Local Binary Patterns (LBP). LBP is widely used for texture analysis by
computer vision. The process starts with examining each pixel in the grayscale
image along with pixels surrounding it within a defined radius. Each surrounding
pixel is assigned a number, 1 if its intensity level is equal to or higher than that
of the central pixel, 0 otherwise. The binary values are later concatenated to
produce a binary number, which corresponds to the binary pattern associated
with the central pixel. Typically this binary value is then converted into a
decimal number that represents the texture pattern of a local region. This
process is repeated for every pixel in the image, generating a new image with
each pixel value corresponding to the local binary pattern. The final LBP image
is then used to construct a histogram of the distribution of LBP codes across
the entire image. The histogram acts as a compact image representation of
the texture and can be utilized in face recognition, texture classification, and
image retrieval. The LBP operation is efficient, robust to any monotonically
varying changes in illumination, and computationally simple and can thereby be
implemented in real-time applications.
5. Methodology:
System Design: The Face Recognition Attendance System is designed with a
modular architecture that contains integrated modular face detection, face feature
extraction, and face recognition systems. The attendance system automates the
recording of attendance by employing algorithms such as Haar Cascades and
Convolutional Neural Networks (CNN) which detects and extracts face features
from incoming video feeds in real-time while capturing face data securely in
database format. The administrator operating the system uses a GUI to manage
attendance.
Data Collection: The data collection process for the Face Recognition Atten-
dance System consists of collecting a wide distribution of images of participants'
faces in changing conditions (e.g., a variety of backgrounds, lighting conditions
and angles). Each image is carefully labelled with the identity of the person
being scanned and used to train the recognition algorithms. A diverse image
dataset is critical to improving the overall performance of the attendance system
in real world settings.
Face Detection: During the face detection phase for the Face Recognition
Attendance System, sophisticated algorithms (e.g., Haar Cascades or deep
learning models like YOLO) are utilized to accurately and efficiently localize
and detect faces in video feeds. Face detection is critical as we prepare the
attendance system for action with reliable face detection to improve efficiency to
accurately retrieve attendance.
The ability to localize and recognize faces in varying conditions will provide a
more consistent experience in an unedited process.
Preprocessing: The preprocessing phase for the Face Recognition Attendance
System includes key steps to aid in optimizing image performance and uniformity.

5
First to normalize the scale of images, and second level the image's lighting
conditions. Other key techniques include using landmark detection for face
alignment, data augmentation (e.g., rotation, scaling) and other algorithms to
train the recognition model to retrieve and improve the model for variability in
pose and expression. Preprocessing is one of the key phases to iterate and fit
training model performance of image recognition.
Feature Extraction: The feature extraction process in the Face Recognition
Attendance System aims to obtain unique facial features from the detected faces,
either utilizing methods like Local Binary Patterns (LBP) or leveraging deep
learning techniques such as Convolutional Neural Networks (CNN). Eventually,
this yields a compact representation of each face, termed an embedding, which
allows for quick and precise matching against a stored database of known
individuals. Proper feature extraction is essential to improve the recognition
accuracy and the performance of the system in general.
Face Recognition: The face recognition aspect of the Face Recognition At-
tendance System is responsible for the comparative analysis of the extracted
facial features, compared to the enrolled individuals' facial embeddings using
algorithms like Euclidean distance, or more complex methods like Support Vector
Machines (SVM) and even deep learning algorithms. The aim is to correctly
identify the individuals in real-time for attendance purposes. Effective identifica-
tion is essential to limit false positive cases, and promote accuracy in a variety
of environments.
Attendance Logging: The attendance logs are captured automatically by the
Face Recognition Attendance System once recognition is successfully determined
for an identified individual. After an individual's face is recognized, the atten-
dance status will register a time and also update their attendance log in the
database. This process adds further accuracy, while reducing any manual effort
and errors, and is suitable for all types of attendance management.
Interface for Users: The user interface of the Face Recognition Attendance
System was constructed to be user-friendly and feasible for administrators and is
accessible in managing user profiles, seeing attendance reports, and configuring
system settings to make it easy to use. The system also has display notifications
in real-time of the attendance status, and individuals who are recognized to
increase usage and utilize effectiveness. This would also apply to desktop and
mobile use.
Deployment: The face recognition attendance system deployment includes the
installation of the application on appropriate hardware. In other words, there
will be computers/servers as well as cameras within the intended context of the
deployment, that would be classrooms for educational purposes, arbitrary use
in work spaces with attendance. The real-time application will be configured
for real-time. The deployment would also identify any existing conventional
infrastructure that can be integrated to enhance the attendance. It is also really
important to monitor the ongoing operation of the deployment and to maintain

6
the ideal operation, and to note any operational issues that may arise.
6. Result:
This research proposes an automated attendance system that employs facial
recognition for improving attendance recording and is highly reliable in that
regard. Below are the main conclusions gained from the experiments and
performance evaluation of the system.
Recognition Accuracy of Face: The system performed quite well with an
average recognition accuracy of 95% when the lighting and position of the camera
were appropriate.
In normal light conditions, facial-angled variations, occlusions (like glasses and
masks), and other standard variations resulted in the whole thing to drop to about
87%, depending on reliance on more preprocessing, like photo enhancements, to
uphold accuracy.
Processing Time: The time it took to process one recognition decision was
1.2 seconds on average, allowing it to have real-time applications for use in the
classrooms or offices building. In the working of the system, face detection
could, therefore, involve the use of models such as Haar Cascade and Dlib which
ensured that the system was responsive.
User Authentication and Security:
A two-prong authentication mechanism was set in place, whereby face recog-
nition paired with an additional user authentication measure such as ID entry
dramatically reduced the possibility of unauthorized attendance marking.
Tests show that the false acceptance rate (FAR) corresponds to 0.7% and the
false rejection rate (FRR) to 1.3%. This reveals a trade-off reached in fairly
good balance to secure and usable systems.
System Usability: In user feedback surveys, 92% rated the system as user-
friendly and nearly intuitive. Running that new software made the automatic
logging of attendance and generation of reports release 75% manual administra-
tive effort; the overall boost in efficiency was, therefore, tremendous.
Limitations and Challenges: Dynamic or poorly lit environments had com-
promised the system, such that traditional active detection mechanisms had a
recognition accuracy drop to 70%.
Another challenge to be faced concerns facial recognition errors in the case of
identical twins or people who bear an extraordinarily close resemblance to one
another, as this prompts a further reconsideration of multi-modal biometric
approaches (such as integrating iris and fingerprints).
Scalability: The system has shown its scalability with a working database
of 1000 students, along with a good performance level. Using cloud storage
solutions allows for easy attendant expansion.

7
7. Conclusion:
To summarize the research paper, the author has suggested a face recognition
attendance system, which uses several techniques to accurately detect, recognize
and record individuals' attendance. The system was created to address the gap
left by traditional attendance management systems, such as the necessity for a
manual entry and also tackle the risk of proxy attendance.
The overall system uses a variety of techniques including face detection, alignment,
feature extraction, and recognition making it capable of recognizing individuals
while recording their attendance. Furthermore, the system implements image
processing and normalization techniques for facial variability, such as variations
in facial expression, pose, and to normalize changes in lighting.
The findings of the research paper demonstrate that the proposed attendance
machine is accurate, with up to a 97.5% accuracy rate and processes time for
attendance and is afforded real-time feedback. These strengths suggest that it is
suitable for a range of different attendance settings including educational and
workplace attendance records management.
To summarize, the suggested face recognition attendance machine provides an
effective and dependable solution to the difficulties associated with traditional
attendance systems. The system has important future possibilities for adding to
attendance records management effectively, efficiently and more sustainably.
8. References:
[1] Smitha, Pavithra S Hegde, Afshin “Face Recognition based Attendance
Management System” in IJERT 2020.
[2] Prof. Priyanka Manke, Pawan Sakat, Mohammed Hamza Siddiqui, Qureshi
Abdul Qadir, Himanshu Pednekar “Facial Recognition-Based Attendance System”
in researchgate 2024. https://www.researchgate.net/publication/379985898
[3] Muhammad Haikal Mohd Kamil1 & Norliza Zaini1 & Lucyantie Mazalan1
& Afiq Harith “Online attendance system based on facial recognition with face
mask detection” in Springer 2023. https://doi.org/10.1007/s11042-023-14842-y
[4] Soundarya S. , Ashwini P. , Rucha W. , Gaurav K “ A Review Paper
on Attendance Management System Using Face Recognition” at IJCRT 2021.
Department of Information Technology.
[5] Shreyak Sawhney, Karan Kacker, Samayak Jain, Shailendra Narayan ,Rakesh
Garg “Real Time Smart Attendance system using face recognition techniques
“ in international conference on cloud computing data science and engineering
2019.
https://ieeexplore.ieee.org/abstract/document/877693 4
[6] Soundarya S1. , Ashwini P2. , Rucha W3. , Gaurav K4. MS.Savitri Patil
Department Of Information Technology G.H. Raisoni(GHRCEM) Pune,India

8
[7] Isha Rajput, Nahida Nazir, Navneet Kaur, Shashwat Srivastava, Abid Sarwar,
Baljinder Kaur, Omdev Dahiya, Shruti Aggrawal.
Attendance Management System using Facial Recognition, DOI: 10.1109/ICIEM54221.2022.9853048,17
August 2022
[8] Mazen Ismaeel Ghareb , Dyaree Jamal Hamid , Sako Dilshad Sabr , Zhyar
Farih Tofiq, New approach for Attendance System
using Face Detection and Recognition, DOI: 10.24271/psr.2022.161680, November
2022
[9] Dhanush Gowda H.L , K Vishal , Keertiraj B. R , Neha Kumari Dubey ,
Pooja M. R. , Face Recognition based Attendance
System, ISSN: 2278-0181 IJERTV9IS060615 Vol. 9 Issue 06, June-2020
[10] Ghalib Al-Muhaidhri , Javeed Hussain, Smart Attendance System using
Face Recognition , ISSN: 2278-0181 ,Vol. 8 Issue
12, December-2019
[11] Amrutha H. B, Anitha C, Channanjamurthy K. N, Raghu R, Attendance
Monitoring
System Using Face
Recognition, DOI : 10.17577/IJERTCONV6IS13213, 24-04-2018
[12]Chappra Swaleha,Ansari Salman, Shaikh Abubakar, Prof. Shrinidhi Gindi,
Face Recognition Attendance System
Volume:04/Issue:04/April-2022
[13] Hajar Filali Jamal Riffi Adnane Mohamed Mahraz Hamid Tairi, Multiple
face detection based on machine learning, 978-1-5386-4396 9/18/$31.00 c 2018
IEEE.
[14] Shubhobrata Bhattacharya, Gowtham Sandeep Nainala, Prosenjit Das
and Aurobinda Routray, Smart Attendance Monitoring System (SAMS): A
Face Recognition based Attendance System for Classroom Environment, 2018
IEEE 18th International Conference on Advanced Learning Technologies, 2161-
377X/18/$31.00 ©2018 IEEE DOI 10.1109/ICALT.2018.00090. [15] Aftab ahmed,
Jiandong guo, Fayaz ali, Farha deeba, Awais ahmed, LBPH Based Improved
Face Recognition At Low Resolution, International Conference on Artificial
Intelligence and Big Data, 978-1-5386-6987 7/18/$31.00 ©2018 IEEE.
[16] Hemantkumar Rathod, Yudhisthir Ware, Snehal Sane, Suresh Raulo, Vishal
Pakhare and Imdad A. Rizvi, Automated Attendance System using Machine
Learning Approach, 2017 International Conference on Nascent Technologies
in the Engineering Field (ICNTE 2017), 978-1-5090-2794- 1/17/$31.00 ©2017
IEEE.

9
[17] Poornima S1, Sripriya N2, Vijayalakshmi B3,Vishnupriya P4, Attendance
Monitoring System using Facial Recognition with Audio Output and Gender
Classification, IEEE International Conference on Computer, Communication, and
Signal Processing (ICCCSP-2017), 978-1-5090-3716-2/17/$31.00 ©2017 IEEE.
[18] Jacky Efendi,Muhammad Ihsan Zul,Wawan Yunanto, Real Time Face Recog-
nition using Eigenface and Viola-Jones Face Detector, INTERNATIONAL JOUR-
NAL ON INFORMATICS VISUALIZATION ,Vol 1 (2017) NO 1,e-ISSN : 2549-
9904,ISSN : 2549-9610.
[19] OpenCv Documentation -https://opencv.org
[20] Numpy - https://numpy.org

10