International Journal of Innovative Research in Computer Science & Technology (IJIRCST)

ISSN: 2347-5552, Volume-11, Issue-1, January 2023

https://doi.org/10.55524/ijircst.2023.11.1.9
Article ID IRP1342, Pages 38-44
www.ijircst.org

Smart Management Attendance System with Facial Recognition

Using Computer Vision Techniques on the Raspberry Pi
Fadhillah Azmi1, Amir Saleh2, and Achmad Ridwan3
1
Department of Electrical Engineering, Universitas Medan Area, Medan, Indonesia
2,3
Department of Informatics Engineering, Universitas Prima Indonesia, Medan, Indonesia
Copyright © 2023 Made to Fadhillah Azmi et al. This is an open-access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

ABSTRACT- In this study, a smart attendance system was recognition for attendance systems can be built using the
created using computer vision techniques embedded in the Multi Task Cascaded Neural Network (MTCNN) technique,
Raspberry Pi device. The initial process is carried out by with results that are practical, reliable, eliminate noise, and
recording students taking certain courses and taking facial save time compared to traditional attendance systems [4].
images for the needs of the system database. In the next Computer vision can be applied to electronic devices to
stage, the system will be regulated according to the time of provide facial detection and recognition in smart attendance
lecture entry to determine which students will attend the system applications that are easy to use and work
lecture. Every student who wants to enter the classroom is automatically by utilizing Internet of Things technology. An
identified by taking facial images with a camera from the attendance system built by applying artificial intelligence has
Raspberry Pi device to identify and determine the time several advantages, namely: an automated time tracking
students enter to attend lectures. Each image taken will be system that is cost-effective, time-saving, and easy to
processed to detect the presence of a face using the Viola- manage [5]. This system is created by evaluating facial
Jones method and to extract features using the LBP method images during the training and testing process to identify
to obtain the feature value of each image. The results which students are present and which students are absent
obtained will be stored in the system for the facial through the images that have been taken and compiling them
recognition process. The final stage of the system being built in the attendance report [6]. The application of this
is to perform face recognition according to the initial image technology can be done on controller devices, such as the
to carry out the attendance process. This process will be Raspberry Pi, and with the help of the OpenCV library as
carried out using the normalized cross correlation (NCC) software for facial recognition [5].
technique, in which the highest feature similarity obtained Systems built with the Raspberry Pi for facial recognition-
between the initial image and the newly captured image is based attendance systems are safer and can verify themselves
the result of recognition by the system. From the trials that automatically. A greater level of recognition will be
have been carried out, the developed system gives good generated from the device by implementing a good facial
results in obtaining attendance management in a fairly recognition algorithm. Individual use of the Raspberry Pi can
efficient manner, and the algorithm proposed for facial enhance attendance management capabilities and function as
recognition obtains good results with an accuracy rate of a standalone piece of hardware [7]. Research conducted for
97.54%. employee attendance management using the Raspberry Pi
has been built, and the developed system will record
KEYWORDS- Smart Attendance System, Face attendance using face matches. After marking attendance,
Recognition, Raspberry Pi, LBP, Normalized Cross names are stored in a file and then sent automatically in real-
Correlation time [8]. In addition, the automatic attendance system helps
in increasing accuracy and speed to achieve real-time
I. INTRODUCTION attendance with high precision and can evaluate attendance
Face recognition plays a very important role in identifying a [9].
person by utilizing their biometric characteristics. Facial A facial recognition-based attendance system can be applied
recognition is a biometric technology that has been used in to manage student attendance and is a solution to reduce the
many fields, such as: security systems, machine-human time needed to calculate attendance, which can be done
interaction, and image processing techniques [1]. The most automatically [1]. Manual attendance management takes
commonly used identification technique is based on facial time away from lectures and teaching staff [10]. The use of
recognition, which takes little time and is very efficient, face-based attendance management is more advantageous
despite being difficult to implement [2]. The facial than RFID and manual techniques. This is because RFID is
recognition process can be done using various techniques, more easily influenced by hackers. In addition, if the RFID
one of which is computer vision. Computer vision is a reader and receiver do not match properly, then lower read
technique used in object recognition that uses various rates occur, resulting in attendance not being processed
approaches so that a person can be properly recognized. properly [2]. An attendance system that is implemented
Several applications of facial recognition can be found in manually in an institution is a tedious process. This is
applications used in smart management attendance system. because the data collection on student attendance in various
Techniques used in facial recognition to build smart departments takes quite a long time. The manual work done
attendance system, such as deep transfer learning, obtain in the maintenance and management of traditional attendance
good recognition results of 98.76% [3]. In addition, facial sheets is difficult [4]. Collecting and managing student

Innovative Research Publication 38

 International Journal of Innovative Research in Computer Science & Technology (IJIRCST)

attendance is one of the most time-consuming jobs in setting the time, and setting the week of meetings. This
schools, universities, and any education system. So, we need process is carried out to run the system in order to obtain
a system that can help process attendance automatically with attendance results for each week. The system is built using a
electronic devices with the help of computer vision display that makes it easy for teachers to manage attendance
techniques. with a website-based application. Meanwhile, the hardware
In this study, a system will be built that can automatically used to record student attendance uses a Raspberry Pi device,
process attendance based on facial recognition using which is supported by computer vision techniques to
techniques from computer vision embedded in the Raspberry recognize each student's face in attendance. The results of the
Pi device. Computer vision is used to perform object introduction will be sent to the database to determine
recognition that works by imitating human visual abilities students who are present, late, and absent in a report that can
[11]. The human brain can recognize objects based on those be imported into the Excel view. The steps taken for facial
that have been seen and store them as initial knowledge. recognition in this study can be seen in Figure 2 below.
When the object is shown again, the brain can respond to it
and classify it based on its resemblance. Computer vision
will imitate this way of working by applying various
algorithms embedded in the Raspberry Pi to carry out the
process of recognizing an object to carry out attendance
management.
The process of developing the system has three important
steps, namely: face detection, face recognition, and
attendance management. Face detection is done to determine
the existence of a face when the image capture process is
carried out. Meanwhile, face recognition is used to identify a
person's face so that it can be recognized by the system.
Meanwhile, attendance management is a data collection
process carried out by the system based on recognized faces,
which records the time the process occurred.
The initial process of the system is carried out by taking all
the facial image data of all students who are used as research
objects and storing them in a database for initial knowledge
of the proposed system. The next stage is the testing phase,
where the system will take an image to mark student
attendance. The system will detect faces, take their images Figure 2: Proposed method for face recognition
using the camera, and compare their features with the initial
image from the previous database. When the identification
In Figure 2, the process is the algorithm used for facial
process is successfully carried out, the system will mark
recognition, which will be uploaded to the Raspberry Pi
attendance and generate a report that can be viewed. In the
device. The Raspberry Pi device is used as a controller and
developed system, the algorithm used for face detection is
information processor for the image that will be used for
Viola-Jones. The detected faces will be cropped and resized
facial recognition for the built-in attendance system. The
and stored as a test dataset for recognition.
block diagram on the Raspberry Pi device used can be seen
in Figure 3 below.
II. METHODOLOGY
This study conducted facial recognition using the normalized
cross correlation method, which was applied to the
Raspberry Pi device to process attendance automatically.
The architecture of the system proposed to build an
intelligent attendance system based on face recognition in
this study can be seen in Figure 1 below.

Figure 3: Raspberry Pi module for face recognition

Figure 1: Smart management attendance system architecture In Figure 3, the Raspberry Pi will detect each facial image
through the connected camera. When there is a face object,
the Raspberry Pi will take the image to be processed and
In Figure 1, the initial process involves registering students,
obtain attribute information. When the information has been

Innovative Research Publication 39

 International Journal of Innovative Research in Computer Science & Technology (IJIRCST)

processed, Raspberry Pi will retrieve the initial image data B. Image Acquisition (Face Image Capture)
from the database to compare its features with the new Acquisition is the process of capturing or taking an image
image. When the image is recognized, the Raspberry Pi will using a tool so that a digital image is obtained. This process
display the data on the display device, record the time taken, uses a tool that can record the image using a digital camera.
and store it in the database for student attendance Image acquisition is carried out after the face has been
management. successfully detected using the Viola-Jones algorithm. This
A. Face Detection process will be carried out during the initial student data
collection and the attendance process. In this study, the initial
Face detection is a necessary first step for a facial recognition
image capture will be carried out by 25 students taking basic
system, where the aim is to localize and explore faces and
electronics courses. The initial image of students can be seen
remove the background [12]. Face detection has also been
in Figure 6 below.
applied to several applications for various needs, such as
image capture and control in public places. There are several
methods that can be used, but generally, use the Viola-Jones
method.
The Viola-Jones algorithm is the most widely applied
method for detecting facial features in real-time with high
performance and accuracy in an image [13]. Viola-Jones is
supported by four main components in its application. The
first component is Haar features, which are used to form a
simple square feature by subtracting the pixel values in the
black area from the pixels in the white area. The second is an
integral image that allows features or characteristics to be
used as detectors, which will calculate very quickly. The
third is AdaBoost machine learning, which is used to select
certain Haar features to use and set threshold values. The
third component is a classifier incorporation method in a
"cascade," which allows the image background area to be
ignored in determining the face area. The process of the
Viola-Jones algorithm can be seen in Figure 4 [12].

Figure 6: Results of initial image taking of students

Haar Integral
Frame
Frame
Features Image C. Databases
Processing After carrying out the process of taking facial images, the
Output
next step is storing the data in the database for facial
Face / Cascade Adaboost Machine
recognition needs in the attendance system. Each image will
Not Face Classifier Learning
be labeled according to the identity number provided by the
campus. The next step is to create the database needed to
collect data on student attendance. The database displays in
Figure 4: Viola-Jones algorithm process the attendance system developed can be seen in Table 1
below.
By using the Viola-Jones algorithm, the image captured by
the camera will be marked and cropped, as shown in Figure Table 1: Database for attendance systems
5 below. No. Name Data Data Types
1 id_att int
2 id_student int
3 week int
4 date date
5 time_in time
6 image_in varchar
7 description varchar
D. Features Extraction
Figure 5: Result of face image detection using the viola- Image feature extraction is an important step in the face
jones algorithm recognition process, which aims to extract biometric feature
values from the image. The feature values obtained are then
used as input for the machine learning model to retrieve the
In Figure 5, these results are obtained by applying the Viola- image information and can be automatically differentiated
Jones algorithm to detect the presence of faces. This [14]. The selection of the best method for obtaining the input
algorithm will carry out the cropping process according to features of an image greatly influences the recognition
the detected face parts and discard the other parts. accuracy when training the algorithm used [15]. One of the
methods in feature extraction is the local binary pattern

Innovative Research Publication 40

 International Journal of Innovative Research in Computer Science & Technology (IJIRCST)

(LBP) method.
Local binary pattern (LBP) is a local spatial structure that
describes an image that has gray invariance and has good
resistance to background noise and visible light [16]. LBP is
used to perform face recognition, where the technique
surrounds each pixel with a central pixel value, labels image
pixels, and generates binary values [17]. The equation used
in finding LBP features uses equations 1 and 2, as follows
[18].
LBPp = ∑Pi=0 s(gi − gc )2i (1)
1 x≥t
s(x) = { (2)
0 x<t
Where: p is the center pixel value, 𝑔𝑐 is the brightness, 𝑔𝑖 is
the brightness of adjacent pixels, and t is the threshold value.
Figure 7: Process of student registration
E. Recognition Process
The similarity of an image can be measured based on the
correlation function using the normalized cross correlation In this study, the data used were those of 25 students who
(NCC) method with a value interval of 0–1, where 1 indicates took basic electronics courses. The results of the registration
the best match [19]. This technique is used for various needs, will then be stored in the database for initial knowledge of
such as facial recognition. NCC has been commonly used as the system's ability to perform facial recognition. Data and
a metric to evaluate the degree of similarity between two facial images that have been stored in the database can be
images being compared. Face recognition using this seen in Figure 8 below.
technique has various advantages and can be done
automatically [20]. The equation for finding the similarity
value between the image data can be stated using equation 3
as follows: [21].
∑𝑁 𝑀
𝑥=1 ∑𝑦=1[𝑎(𝑥,𝑦).𝑏(𝑥,𝑦)]
𝑁𝐶𝐶 = (3)
√(∑𝑁 𝑀 2 𝑁 𝑀 2
𝑥 ∑𝑦 [𝑎(𝑥,𝑦)] ).√(∑𝑥 ∑𝑦 [𝑏(𝑥,𝑦)] )

Information:
a : The first image features value
b : The second image features value
N : Number of record data
M : Number of column data
By using equation 3, each query image features will be
compared with the database images to obtain the highest data
similarity. The highest similarity result is the result of facial
recognition, which will be stored in the system. This method
was chosen because of its simplicity, because it can be used
in image retrieval systems, and because it obtains good Figure 8: Registration results and facial image storage
accuracy in face recognition [22].
F. Saving Attendance In Figure 8, the system will be given an initial image for the
The results of the recognized image will be stored in the face recognition process in the smart attendance system. The
database to determine the student's attendance time. This will image will have attribute values calculated using the LBP
make it easier for lecturers who want to record students who method. This image will be used as a reference in carrying
attend lectures and save time. out the identification process, which is also called a database
image. After registering, the next step is to set a lecture
schedule. The display settings can be seen in Figure 9 below.
III. RESULT AND DISCUSSION
A. Results of System Implementation
The interface on the system is built using a website
application to make it easier to manage student attendance.
The initial stage of the system is carried out by registering
each student who attends lectures. The data needed is in the
form of name, student ID, email, gender, and a photo taken
as a database image for the facial recognition process. The
registration interface displays of the system being developed
can be seen in Figure 7 below.

Innovative Research Publication 41

 International Journal of Innovative Research in Computer Science & Technology (IJIRCST)

when the image is captured and recognized by the system.

When a student exceeds the specified time threshold, or in
other words, exceeds the entry time, the system will record
that the student is late, and the LCD will display "You're
late!" "Please be more disciplined."
The attendance system built will be used each week
according to the lecture schedule. In this study, the system
will record the results of student attendance for 12 weeks,
and each week the attendance results will be stored in the
database. You can view the report by selecting the week of
the meeting. The display of student attendance can be seen in
Figure 11 below.

Figure 9: Creating a lecture schedule

In Figure 9, the system will be arranged according to the

lecture schedule, such as the class schedule, date, and time of
entry. When students want to enter the classroom, each
student must do a face scan using the device that was
developed. When the device detects a face, the system will
mark and crop the image, and the recognition process will be
carried out. The image that has been captured will be
compared for similarity with the database image for face
matching using the NCC method. The image with the highest
similarity measure between them will be used as the result of
recognition, and the system will automatically record the
time the image was taken and store it in the database for
display. The results of testing the attendance system can be Figure 11: Results of student attendance reports
seen in Figure 10 below.
After selecting the week of the meeting, the system will
display a graph to show the percentage of student attendance,
as shown in Figure 12 below.

Figure 10: Student attendance system test results

In Figure 10, the system will determine the results of face

matching, which is done when you want to enter lectures Figure 12: Percentage of student attendance
with the previous database image. After performing face
scanning and face matching, the system will record student After the lecture is finished, the teacher can import the
attendance when it has successfully identified them. The data attendance results that were previously recorded in Excel
specified in the built system is classified into three format. This is done to find out which students are present,
categories: present, late, and absent. "Present" is a late, and absent. The results of import attendance from the
description that appears when the attendance process is developed system can be seen in Figure 13 below.
carried out according to the time of lecture entry with a delay
tolerance of 10 minutes. "Late" is a statement that appears
when a student is absent beyond the specified time limit.
When students fail to perform facial recognition or are late
for more than 30 minutes, While Absent will appear. Besides
that, there is also an indicator on the LCD with the caption
"Attendance Successfully Done!" This process can be done

Innovative Research Publication 42

 International Journal of Innovative Research in Computer Science & Technology (IJIRCST)

will be calculated using the NCC method, and the highest

value of the equation will be used as the result of the
recognition made by the system. The stages of face
recognition can be illustrated in Figure 15 below.

Figure 15: Face recognition results

Figure 13: Import of student attendance data results
Based on Figure 15, the highest level of similarity is worth 1
As for the total number of students in attendance as a whole, and is the result of recognition from the attendance system
it can be seen using the graph in Figure 14 below. that was built. This process will be carried out in 12 meetings
with a total of 285 query images, and the results of the
introduction that has been carried out for each student can be
12 seen in Table 2 below.
11
10 Table 2: Face recognition results using the proposed
method
9
Total Correct Incorrect
8 No. Student ID
Attendance Number Number
7 1 218120002 12 12 0
6 2 218120003 12 11 1
3 218120004 12 11 1
5 4 218120007 10 10 0
1 3 5 7 9 11 13 15 17 19 21 23 25 5 218120009 12 11 1
6 218120010 12 12 0
Figure 14: Results of student attendance 7 218120011 10 10 0
8 218120012 12 11 1
9 218120013 11 11 0
Overall, the proposed system performs well in terms of 10 218120014 12 11 1
managing student attendance. This system can record student 11 218120017 11 11 0
attendance for each week and report the total attendance 12 218120018 12 12 0
according to the class schedule. 13 218120019 11 11 0
14 218120020 11 11 0
B. Face Recognition Accuracy Testing 15 218120021 12 10 2
This research was developed for a smart attendance system 16 218120022 11 11 0
using computer vision techniques embedded in the 17 218120024 12 12 0
Raspberry Pi device as the hardware used for the detection 18 218120026 12 12 0
process, image processing, and facial recognition process. 19 218120029 11 11 0
All computing processes will be performed on the device, 20 218120031 10 10 0
and the results will be sent to the database system. Every face 21 218120034 12 12 0
captured by the Raspberry Pi camera will immediately 22 218120035 11 11 0
23 218120037 12 12 0
undergo a face detection process and produce an image in the
24 218120038 11 11 0
form of a snippet of the detection results from the Viola-
25 218120041 11 11 0
Jones method. The image results will be used as an image Total 285 278 7
query and for feature extraction using the LBP method.
The next step is to calculate all the images in the database
In Table 2, the face recognition process was carried out by
(initial image) for the value of the LBP feature and compare 25 students during the attendance process. The total testing
the closeness of the value with the query image. This value was carried out 12 times, so the images used to carry out the

Innovative Research Publication 43

 International Journal of Innovative Research in Computer Science & Technology (IJIRCST)

recognition process were 285 images. From these results, recognition,” J. Adv. Res. Dyn. Control Syst., vol. 12, no. 5
there are 278 images that are well recognized. The total facial Special Issue, pp. 1079–1084, 2020, doi:
recognition accuracy with the proposed method is 97.54%. 10.5373/JARDCS/V12SP5/20201860.
Some face recognition is not carried out correctly, which is [7] T. Menezes, “Face Recognition Attendance System using
Raspberry Pi,” Int. J. Res. Appl. Sci. Eng. Technol., vol. 9, no. 8,
influenced by various conditions, such as the brightness level pp. 1145–1149, 2021, doi: 10.22214/ijraset.2021.37499.
during the retrieval process, and others, so that the attendance [8] A. K. L, I. V, C. A, R. PauL, and S. Banerjee, “Smart Attendance
filling process is carried out manually. Management System Using Raspberry Pi and Deep Learning
From the overall test results of the system being built, the Technique,” 2022, doi: 10.4108/eai.7-12-2021.2315121.
facial recognition-based attendance system can be carried out [9] N. R and D. K. P, “Student Smart Attendance Through Face
well with a recognition result of 97.54%. This system can be Recognition using Machine Learning Algorithm,” Int. J. Recent
implemented effectively for managing attendance and works Technol. Eng., vol. 9, no. 1, pp. 2348–2352, 2020, doi:
automatically. In addition, the system developed makes it 10.35940/ijrte.a2927.059120.
easy to record and process student attendance as a condition [10] A. J. Moshayedi, A. S. Roy, L. liao, M. Gheisari, A. A. Abbasi,
and S. M. H. Bamakan, “Automation Attendance Systems
for conducting an assessment process for lectures that have Approaches: A Practical Review,” BOHR Int. J. Internet Things
been conducted. Res., vol. 1, no. 1, pp. 7–15, 2021, doi: 10.54646/bijiotr.003.
[11] T. A. Dompeipen, M. E. I. Najoan, J. T. Elektro, U. Sam, and R.
IV. CONCLUSION Manado, “SSD, Mobile-net,” vol. 16, no. 1, pp. 65–76, 2021.
[12] I. Gusti Ngurah Made Kris Raya, A. N. Jati, and R. E. Saputra,
Based on the results of the attendance system developed to “Analysis realization of Viola-Jones method for face detection on
achieve good results, student attendance management can be CCTV camera based on embedded system,” Proc. 2017 Int.
done easily, and attendance reports can be obtained Conf. Robot. Biomimetics, Intell. Comput. Syst. Robionetics
automatically for each week. In addition, the proposed 2017, vol. 2017-Decem, pp. 1–5, 2017, doi:
system provides a good level of success or accuracy in 10.1109/ROBIONETICS.2017.8203427.
performing facial recognition for the attendance process of [13] J. Efendi, M. I. Zul, and W. Yunanto, “Real time face recognition
97.54%. These results indicate that the proposed system for using eigenface and viola-jones face detector,” Int. J. Informatics
Vis., vol. 1, no. 1, pp. 16–22, 2017, doi: 10.30630/joiv.1.1.15.
conducting attendance management provides better results
[14] S. Bakheet and A. Al-Hamadi, “Automatic detection of COVID-
than the traditional attendance system. 19 using pruned GLCM-Based texture features and LDCRF
classification,” Comput. Biol. Med., vol. 137, no. August, p.
V. FUTURE SCOPE 104781, 2021, doi: 10.1016/j.compbiomed.2021.104781.
[15] Priyanka and D. Kumar, “Feature Extraction and Selection of
The problems found in testing the developed system include kidney Ultrasound Images Using GLCM and PCA,” Procedia
the inability to perform facial recognition on two or more Comput. Sci., vol. 167, no. 2019, pp. 1722–1731, 2020, doi:
students at the same time. The development of this research 10.1016/j.procs.2020.03.382.
that might be done is to develop an algorithm that can [16] Q. Zhang, “Facial expression recognition in VGG network based
perform facial recognition for two or more students at once, on LBP feature extraction,” Proc. - 2020 5th Int. Conf. Mech.
or in other words, multiface recognition. This method may Control Comput. Eng. ICMCCE 2020, pp. 2089–2092, 2020, doi:
be more difficult to do using a facial recognition algorithm, 10.1109/ICMCCE51767.2020.00454.
S. J. Elias et al., “Face recognition attendance system using local
but it is an effective one in terms of time and saves image [17]
binary pattern (LBP),” Bull. Electr. Eng. Informatics, vol. 8, no.
data storage for the facial recognition process. 1, pp. 239–245, 2019, doi: 10.11591/eei.v8i1.1439.
[18] L. Zhang, B. Zhong, and A. Yang, “Building Change Detection
CONFLICTS OF INTEREST using Object-Oriented LBP Feature Map in Very High Spatial
Resolution Imagery,” 2019 10th Int. Work. Anal. Multitemporal
The authors declare that they have no conflicts of interest. Remote Sens. Images, MultiTemp 2019, pp. 1–4, 2019, doi:
10.1109/Multi-Temp.2019.8866919.
REFERENCES [19] A. Nakhmani and A. Tannenbaum, “A New Distance Measure
Based on Generalized Image Normalized Cross-Correlation for
[1] M. Kasiselvanathan, A. Kalaiselvi, S. P. Vimal, and V. Robust Video Tracking and Image Recognition,” Pattern
Sangeetha, “Smart Attendance Management System Based On Recognit Lett., 2012, doi: 10.1016/j.patrec.2012.10.025.A.
Face Recognition Algorithm Engineering, Communication,” [20] N. T. Abdulsada and S. M. Ali, “Automatic Face Recognition
Int. J. Pure Appl. Math., vol. 120, no. 5, pp. 1377–1384, 2018, using Normalized Cross Correlation (NCC) Function with
doi: https://acadpubl.eu/hub/2018-120-5/5/531.pdf. Variable Template Size,” AIP Conf. Proc., vol. 2437, no. August,
[2] K. R. Kiran and S. Mekala, “Face Recognition Attendance 2022, doi: 10.1063/5.0093157.
System using Raspberry Pi,” Int. J. Pure Appl. Math., vol. 118, [21] A. Kaso, “Computation of the normalized cross-correlation by
no. 20, pp. 3061–3065, 2018, doi: fast Fourier transform,” PLoS One, 2018, doi:
https://acadpubl.eu/hub/2018-118-21/articles/21d/19.pdf. 10.1371/journal.pone.0203434.
[3] K. Alhanaee, M. Alhammadi, N. Almenhali, and M. Shatnawi, [22] A. Saleh, D. Suryandy, and J. Nainggolan, “Face Image Retrieval
“Face recognition smart attendance system using deep transfer System Using Combination Method of Self Organizing Map and
learning,” Procedia Comput. Sci., vol. 192, pp. 4093–4102, Normalized Cross Correlation,” J. Infokum, vol. 9, no. 2, pp.
2021, doi: 10.1016/j.procs.2021.09.184. 219–228, 2021.
[4] G. Anitha, P. S. Devi, J. V. Sri, and D. Priyanka, “Face
Recognition Based Attendance System Using Mtcnn and
Facenet,” Zeichen, vol. 6, no. 8, pp. 189–195, 2020.
[5] A. S. Nadhan et al., “Smart Attendance Monitoring Technology
for Industry 4.0,” J. Nanomater., vol. 2022, pp. 1–9, 2022, doi:
10.1155/2022/4899768.
[6] R. S. Sabeenian, S. Aravind, P. Arunkumar, P. Harrish Joshua,
and G. Eswarraj, “Smart attendance system using face

Innovative Research Publication 44