A Review Paper on Real Time face recognition class based

Attendance System
Zain ul Abideen Qureshi
Department of Electrical Engineering Faculty of
Electronics and Electrical Engineering University of
Engineering & Technology Taxila, Pakistan
qzain06@gmail.com

Abstract
Providing accurate attendance marking system in real-time is challenging. It is tough to
mark the attendance of a student in the large classroom when there are many students
attending the class. Many attendance management systems have been implemented in
the recent research. However, the attendance management system based on facial
recognition still has issues. Thus many research have been conducted to improve
system. This paper reviewed the previous works on attendance management system
based on facial recognition. This article does not only provide the literature review on
the earlier work or related work, but it also provides the deep analysis of Principal
Component Analysis, discussion, suggestions for future work.
Keywords
Principal Component Analysis, Facial Recognition, Histogram Oriented Gradient,
Automatic System, AMS.

I. INTRODUCTION
Facial recognition development has received many interests in recent years. It is a
critical application in image analysis yet it is very challenging to create an automated
system based on facial recognition, a system with ability to recognize human face
accurately. One application of facial recognition is in the field of attendance
management system. The manual attendance system is time-consuming, thus many
research has been conducted with the automatic [1], [2], [3] or smart attendance
management [4] system to resolve this issue. One solution is the application of
biometric attendance management system. However, it is difficult to verify each student
in the classroom as there are many students who attend the class, and if the system
cannot detect or recognize one student, it will interrupt the learning process.
In addition, the biometric system needs much hardware that requires high cost and a lot
of interaction with the students that makes it a time-consuming system. The research on
applying real-time facial recognition in attendance management system has been a real
challenge. Automatic attendance marking can solve the main issues such as the error
when inputting the data from the sheet to the manual attendance system, especially
there is a concern that the data is not 100% accurate in the traditional attendance
marking method considering the extremely large number of students in a university.
The facial recognition process has many steps such as capture, extraction, comparison,
and matching [5]. The operation in each process is:

1. Step one

Capture is the way to snap the picture during the enrollment of the system.

2. Second step

Extraction is used for finding or extract the specific feature from the face.

3. Third step

Third step is comparison, where new input is used for comparison with the
database (sample data).

4. Fourth step
 Fourth step is matching the system will try to find the matching of the new face
with the registered face based on extraction and comparison process. The
sequence of the processes is illustrated in Fig. 1.

Fig. 1. The operate process of face recognition system

There are several methods and algorithms that can be implemented in facial recognition
system. Study conducted by [6] implements LBP (local binary pattern) to recognize the
human face and SVM (support vector machine) to find the similarities of the images
stored in the database with the image acquired by camera inside the classroom. There
are a lot of limiting on the research done in the past. The detail of the limitations will be
introduced in next part of literature survey. The literature review gives a summaries of
information about the previous research and how each researcher implements the
attendance management system by using face recognition. In section five gives a
comparison of each paper based on the criteria by the author and provides the small
discussion about face recognition algorithm in attendance management system. At the
last section, will concludes all the discussion in this paper and gives some advice for
refinement of attendance management system.

II. FACE RECOGNITION

In the introduction, an overview of facial recognition in attendance marking system is
discussed. Then, in this section, we explain the various approaches to facial recognition
such as:

A. Geometric Approach
This approach recognizes human based on geometry point on the face.
There are many features of geometric approach that has been explored: [7]
uses geometric approach to improve the accuracy of the system when the
images have complex background.

B. PCA Based
PCA is a statistical method used in facial recognition for feature extraction
and used to remove redundant information. PCA also used to decrease
dimensionality and conserve the information. When new image is input to the
system; facial recognition will process it to find the match between the new
image and the image inside the databases. Principal component or Eigen
face is a face image that contains feature and characteristic. The image with
the M ×M can be written to 1-dimensional M2 to reduce the dimensionality.

C. LDA Based
This method is used to maximize the scatter image in the same or different
class, which is enhanced from PCA. In PCA, a subspace is selected that
contains the most variation. Since 1997, Fisher faces [8] has been used in
the application of Fishers linear discriminant (FLD). The Fishers linear
discriminant chooses the linear subspace Φ that maximize the ratio.

T
Φ Sb Φ
T
Φ Sw Φ (1)

Where,
m
sb =∑ ¿ ( x i−x ) ( x i−x )
T
(2)
i=1

(3)
m
T
sw =∑ ∑ ( x i−x )(xi −x)
i=1 x ∈ Xi

Sb : between-class scatter
Sw : within-class scatter
 m : number of subject in the database

In [8], this method was tested with a front-view of face images, and the result had less
error than Eigen faces-based method.
D. Local Binary Pattern
This approach is used in computer vision, which is first introduced in 1990 for
classification. In 1994, research papers [9] and [10] showed that combination
of LBP and HOG (histogram of oriented gradient) had a great performance in
texture classification and it improved the performance detection on some
relevant data sets.

E. Active Appearance Model

This model was first introduced by Edwards, Cootes and Taylor [11], [12] in
the same year. This approach has been used extensively to do facial
matching and it is also used in medical interpretation field.

III. FACE RECOGNITION TECHNIQUES

A. Principal Component Analysis (PCA)

PCA is commonly known as method of dimension reduction and feature
selection. However, it is also used to extract the number of principal component
of multidimensional data. Human recognition system using principal component
analysis was first developed by Turk and Pentland [13] which was used to solve
facial recognition in two-dimensional rather than three-dimensional geometry
because both of the captured image and basic vector constructed by the PCA
have the same dimension [14]. Reconstruction of the human face has been done
by [15] for solving the retrieval of large database images. In the recent
development of automatic facial recognition, the feature extraction has become
significant to extract features such as nose, eyes, and mouth. Moreover, the
main goal of PCA is keeping the original information of the data, minimizing
original loss information and improving the face analysis. The principal
component analysis mainly uses Eigen face that has extremely large size, thus it
requires dimension reduction process [16]. Study conducted by [17] introduces a
way to calculate the discriminant feature of PCA by using a recursive algorithm.
In the following section focus on the step of PCA in facial recognition system by
using Eigen face.

 Capturing and preparing image for training Suppose we have images X1,
X2, ..., X2 for training in face recognition system. The images must be the
same size.
 All the face in Xi are transformed as a vector and put into a training set.

S = {Γi, ..., Γn} (4)

Which we assume images has the size N × N or 1× N2 dimensional.

 Computing the mean face

i=n
1 (5)
X= ∑x
N i=1 n

 Computing the average of face vector

The mean face is subtracted by original face and the result will be stored in
variable.

Φi = Γi − X (6)

 Computing the covariance

Calculate the covariance matrix with the following mathematic formula

n
(7)
∑ (Γ i−X )(Γ i −X )T
C= i=1
n

 Calculate the covariance matrix of Eigenvector and Eigenvalues

All new face is transformed into its Eigen face components and the
resulting weights from the weight vectors with:
ω= weight
µ = eigenvector
Γ = new input images vector
X¯ = mean face The weight vector Ω T is given by Ω

T = {ω1, ω2, ..., ωn} (8)

 Then Eigen faces handle face recognition - The Training data project into
PCA subspace - Query image project into PCA subspace - Find the
nearest neighbor between the database image and training image which
already projects into PCA subspace.

IV. LITERATURE SURVEY

The primary purpose of this review paper was to find the solutions
provided by other authors and consider the limitations of their proposed
methods. The best solution then proposed after reviewing all methods. In
the Fig. 2 show the current research paper has been published [18], [19]
[6], [20], [21], [22], [1].

In 2005, Kawaguchi [23] introduces a lecture attendance system with a

new method called continuous monitoring, with the students attendance
marked automatically by the camera which captures the photo of a
student in the class. The system has simple architecture with only using
two cameras installed on the wall of the class. The first camera is a
capturing camera used to capture the student image in the class and the
second camera is sensor camera used to get the seat of a student inside
the class and the capturing camera will snap the student image. The
system then compares the picture taking from the capturing camera and
images in the database. This process is done repeatedly to complete the
attendance marking process.

In 2010 V. Shehu and A. Dika [2] proposed real-time computer vision

algorithm in automatic attendance management system. The system uses
non-intrusive camera that can capture images in the classroom and
compares the extracted face from the captured image with database
inside the system. This system also uses machine learning algorithm
which is usually used in computer vision. In addition, Haar classifier is
used to train the images from the camera. The face captured by the
camera then converted to grayscale and the image is put to subtraction
process. The image then stored on the server to be processed later.

In 2012, N. Kar [24] introduced an automated attendance management

system using facial recognition technique that used the principal
component analysis. This system uses two libraries which are OpenCV, a
computer vision library, and FLTK (light tool kit). Both libraries help the
development of the system, for example OpenCV supports algorithm [20]
and FLTK [21] is used to design the interface. In this system, there are
two processes, namely request matching and adding new face to
database.
In request matching, the first step is opening the camera and capturing
the photo, then the face is extracted from the image. The next step is
recognizing the face with the training data and projecting the extracted
face onto the principal component analysis. The final step is displaying
the face that closely matched the acquired image. Meanwhile, adding
new face to database process is started with capturing the photo, then the
face is extracted from the image. The Haar cascade method then
performed to the image to find the object in the image in different window
size. The next step is storing the image into the database, then learning
the face, and followed with application of principal component analysis
algorithm. The final step is storing the information inside the face XML
file. The system is focused on the algorithm to improve the face detection
from acquired images or videos.

In 2013, J. Joseph and K. P. Zacharia [3] proposes a system which

implements automatic attendance using facial recognition. The system
can extract the object in the face such as nose or mouth by using
MATLAB with principal component analysis (PCA). The system [7]
designed to resolve the issues of attendance marking system such as the
time-consuming issue. The study result shows that the system can
recognize face image in the dark background or different view of the face
in the classroom.

In 2014, Jyotshana Kanti and A. Papola [4] proposes a smart attendance

marking system that combines two different algorithms, principal
component analysis and artificial neural network. The study is able to
solve the traditional attendance marking system and resolve the time-
consuming issue. Principal component analysis extracts and identifies the
similarities of the face database and acquired images in the system.
Meanwhile, artificial neural network is used to solve the problem or learn
from the input data and the expected value. This system also uses back
propagation algorithm combined with mathematical function. The result
shows that the system is able to recognize faces in various environment.
The evolution of attendance management system using face recognition system can be
shown by the figure
 Fig. 2. The evolution of attendance management system using face recognition

In [22], Priyanka Thakare proposes a method using Eigen face and principal component
analysis with architecture as follows. The camera is installed in the front, which is used
to capture entire face of the students inside the class. Then, the captured images are
transferred into the system as inputs. The images captured from the camera could be
too dark or too bright, thus enhancement is needed to convert them to gray images. In
the next step, histogram normalization is used to remove the contrast of the images,
thus it is easy to recognize the students who sit in the back row. The median filter is
used to remove noise from the images. Noise sometimes still occurs even when high
definition camera is used. The system also implements skin classification that changes
all pixels to black, except the pixels that are close to the skin.
 V. COMPARISON OF FACE RECOGNITION TECHNIQUES

The aforementioned papers proposes several algorithms. Table I covers six papers that
are related to facial recognition area and an overview of all papers, including the
problems and cover the limitations of each paper used algorithms. We also give a
summary of every papers.
In Table II gives a brief pros and cons of each paper. All papers have conducted with
attendance management system should be improved based on each paper in Table III.
Which each paper has their own pros and cons on difference environment of the
implementation system. In the artificial intelligence-based method, artificial neural
network is able to classify face images successfully and it has the best system
performance compared to other artificial intelligence based or conventional algorithms.
Furthermore, integration of traditional and artificial intelligence-based algorithm such as
presented paper titled Face Recognition using Principal Component Analysis and
Neural Network also achieved better accuracy and performance than previous
attendance management based facial recognition system. The various algorithms
discussed in other papers are principal component analysis with AdaBoost [26], PCA
with error back-propagation [27] [28], and PCA with ANN and Wavelet [29]. Each paper
shows that the system accuracy is higher than the system with PCA that is not
incorporated with ANN.

TABLE I
SUMMARY PAPER

Authors Problem Accuracy Cost Summaries

Visar Shehu [2] The recognition rate Using HAAR
is 56%, having a Yes high Classifier and
problem to recognize computer vision
student in year 3 or 4 algorithm to
implement face
recognition
NAVAZ [25] Low accuracy with Using PCA to
the big size of images Yes High train and reduce
to train with PCA dimensionality
and ANN to
classify input
data and find
the pattern
N.Kar [24] Repeat image Using
 capturing Low High Eigenvector and
Eigenvalue for
face recognition
Joseph [3] Validation of the Using PCA with
student once marked Yes High MATLAB to
present is not done implement face
recognition
E.Reakha [18] Recognition only Using PCA and
frontal face Low High Eigen face to do
a better
P.Wagh [22] Low accuracy in Using PCA and
lighting Low High using Histogram,
Remove Noise

TABLE II
THE PROS AND CONS OF EACH PAPER

Authors Pros Cons

Visar Shehu [2] The system integrated with the existing The ability of face
system that can use the existing feature detection still lacks on the
from Learning Management System system. The existing face
database need to be
updated
NAVAZ [25] The accuracy is high due to the The computation time and
combination PCA with ANN. ANN use implementation cost is
for classificaton is more accuracy than higher because of combine
PCA with Eigenface PCA with ANN
N.Kar [24] The system have prevented the fake The recognition task is high
attendance due to the implementation only in frontal face. In the
clock time in and time, which used for direction of 54° the
checking the student presence inside accuracy only 58%
the class for the period or not
Joseph [3] The pre-processing cropping the region
The system have tested
of interest that make the accuracy with the single images only
more higher. based on the result.
E.Reakha [18] The pre-processing cropping the region
The author implmeneted
of interest that make the accuracy only PCA and Eigenface.
more higher. Due to the algorihtm
implmeent in they system,
so the computational cost
is high.
P.Wagh [22] Providing a better accuracy by This will be a fake
implementation Ada-Boost algorithm attendance if the student
on Face Detection. The database is just come in classroom and
update to date with the new images. then go back before the
Skin classification use for improving the end of class period
accuracy of face detection

VI. System Progress

For the student details, pictures of every student are taken through the webcam or any
other camera available this is done by capturing the real-time video. It contains 60
samples that are stored along with the student id in a student detail folder. Unknown
images are stored in the folder name unknown images. The real-time face recognition
the student’s face is highlighted and detected on the window then detected faces are
cropped and stored in the respective student's name folder. The whole process is done
in student registration flow.
 Figure 1-Student Registration Flow

Feature Extraction
Feature extraction is used to extract the face features such as eyes, edge features, line
features, and center-surrounded features. Extraction of facial features for comparison
from the training set and test image for face detection. After the extraction of face
features the data is stored in the files that are classifier.it is created to store the
information. Student's images with Id are stored in the same folder.

Figure 2-Feature Extraction

Generating Participation Progress

This work is being completed in five phases:

Stage 1:
a) Data set
Generating Data for Training Initially the framework is prepared that is the edited
pictures are spared to the database and they experience location and acknowledgment.
Further, this information will be used to think about the distinguished pictures in all the
transferred records and imprint the participation.

Stage 2:
b) Capturing
In this the catching of the video or picture will be finished utilizing a gadget and the
caught record is transferred to the server utilizing web application.

Stage 3:
c) Face Detection
The record transferred to the server experiences face identification utilizing Viola what's
more, Jones calculation. The casings got are checked for the countenances and those
are trimmed for further acknowledgment.

Stage 4:
d) Face Recognition
The recognized pictures experience a relationship with the prepared pictures of the
databases. By this, the identified pictures are currently perceived.

Stage 5:
e) Attendance Marking
After the acknowledgment procedure the understudies perceived are looked in the
database and their participation is checked.
Database Design

This is the detail flow diagram of the project which involves the following steps.

1. Take images

Gathering data take pictures through the webcam or any other camera available. This is
all done through open cv.it is an open-source computer vision library used for image
recognition.

2.Training

After taking the image sample we have to click Train Image button. Now it takes a few
seconds to train the machine for the images that are taken by clicking the Take Image
button and creates a Trainner.yml file and store in the Training Image Label folder.
3. Track image

By clicking the Track Image button camera of the running machine is opened again. If
the face is recognized by system then Id and Name of person are shown on Image.

4. Attendance

Press Q(or q) for quit this window. After quitting it attendance of person will be stored in
the Attendance folder as CSV file with name, id, date and time and it is also available in
the window. The whole process is shown in the figure
Background Level

Students details are entered into automatic attendance management system and then
the process will start .first it will enable the camera and then capture the image and then
it will extract the face features and send to the known image file these are all stored in
the name of generated files when the features will extracted it will recognize the
students with the known matching data provided by the data set. Then further reports
are delivered to faculty after that, the reports are checked by the management and
further details are send to the authorized authority. Following are the flow diagram that
shows how the whole process will flow and managed the data.

Figure 3- DFD- Background Level

System Screenshots and Explanation

People perform face acknowledgment naturally consistently and for all intents and
purposes with no exertion. There is different kind of face recognition algorithm. Every
technique has an alternate way to deal with concentrate the picture data and play out
the coordinating with the info picture. Today we going to discuss one of the most
established and increasingly prevalent face acknowledgment calculations: Local Binary
Patterns Histograms. (LBPH)
Utilizing the LBP joined with histograms we can speak to the face pictures with a basic
information vector. As LBP is a visual descriptor it can likewise be utilized for face
acknowledgment undertakings, as can be found in the accompanying bit by bit
clarification

Steps
Since we discover somewhat more about face acknowledgment and the LBPH, we
should go further and see the means of the calculation:
Parameters: the LBPH utilizes 4 parameters:
Radius: The radius is utilized to construct the round nearby double design and speaks
to the range around the focal pixel. It is normally set to 1.
Neighbors: The quantity of test focuses to manufacture the roundabout nearby parallel
example. Remember: the more example focuses you incorporate, the higher the
computational expense. It is normally set to 8.
Grid X: The number of cells in the flat course. The more cells, the better the framework,
the higher the dimensionality of the subsequent component vector. It is generally set to
8.
Grid Y: The number of cells in the vertical bearing. The more cells, the better the matrix,
the higher the dimensionality of the subsequent component vector. It is typically set to 8.
When we Run the program in pychram editor a window is open which contain the
following terms such as
1. Enter id
2. Enter Name
3. Notification
4. Take images
5. Train images
6. Track images
7. Quit
8. Attendance
Take images

After inter the details of the student id and name the first step is to take images from the
webcam or any other camera install in the classroom. This is all done through open cv.it
is an open-source computer vision library used for image recognition.
Train images

After taking the image sample we have to click the Train Image button. Now it takes a
few seconds to train the machine for the images that are taken by clicking the Take
Image button and creates a Trainner.yml file and store in the Training Image Label
folder. Harcascade classifier detects the desired portion of the image captured as in this
project the desired portion is our face. We are using a frontal face detection haar
cascade classifier. We are storing these images in a folder name training image.
Unknown images like glass appear in the video or any other thing if captured stored in
the unknown image folder.
Track images

By clicking the Track Image button camera of the running machine is opened again. If
the data is being in the folder which we take the images from the camera so when we
click on the track images it will start recognizing the data from the previous data in the
folder. If a face is recognized by the system then Id and Name of person are shown on
Image.
Attendance

Press Q(or q) for quit this window. After quitting it attendance of person will be stored in
the Attendance folder as CSV file with name, id, date and time and it is also available in
the window. The following images are the details of the whole process.

VII. CONCLUSION

This paper reviews several methods in terms of overall system capacity, throughput,
and accuracy. The result shows that the PCA algorithm is incredibly effective in
extensive database. PCA has better performance in the system of attendance
management based on facial recognition than manual attendance system that is time-
consuming. Convolutional neural network also contributes to attendance management
system based on facial recognition by providing strong classifier The focus in future
work is improving the accuracy of the system by incorporating principal component
analysis with convolutional neural network. The objective is to obtain good
generalization abilities. In [25], ANN and PCA has been integrated to solve a blocking
issue of attendance management system based on facial recognition. However, this
system still has issues with system performance and accuracy in recognizing human
face.
 Future work will use fast PCA with back-propagation to resolve this problem.
Our project is recognizing marking the attendance of students whose data is stored
already .future work can be that like with the passage of time changes occur in the
human faces and body so the shape and size of face is change for this purpose we
need an intelligent algorithm that marks the attendance of students and as well as
stored the changing feature of face and stored it in the particular student record time to
time.
To avoid from terrorist activities in the institution it will help in a way that if an unknown
person comes in class an algorithm is made so efficient that when it detects the
unknown face a buzzer connected to the system start sounding.

REFERENCES

[1] A. Ahmedi and S. Nandyal, “An Automatic Attendance System Using Image
processing,” pp. 1–8, 2015.

[2] V. Shehu and A. Dika, “Using real time computer vision algorithms in automatic
attendance management systems,” Information Technology Interfaces (ITI),
2010 32nd International Conference, pp. 397–402, 2010.

[3] J. Joseph and K. P. Zacharia, “Automatic Attendance Management System

Using Face Recognition,” International Journal of Science and Research (IJSR),
vol. 2, no. 11, pp. 327–330, 2013.
[4] J. Kanti and A. Papola, “Smart Attendance using Face Recognition with
Percentage Analyzer,” vol. 3, no. 6, pp. 7321–7324, 2014

[5] T. H. Le, “Applying Artificial Neural Networks for Face Recognition,” Advances in
Artificial Neural Systems, 2011.

[6] P. Mehta, “An Efficient Attendance Management Sytem based on Face

Recognition using Matlab and Raspberry Pi 2,” International Journal of
Engineering Technology Science and Research IJETSR, vol. 3, no. 5, pp. 71–
78, 2016.
[7] S.-h. Jeng, H. Y. M. Liao, C. C. Han, M. Y. Chern, and Y. T. Liu, “Facial feature
detection using geometrical face model: An efficient approach,” Pattern
Recognition, vol. 31, no. 3, pp. 273–282, 1998.
[8] P. N. Belhumeur, J. P. Hespanha, and D. J. Kriegman, “Eigenfaces vs.
fisherfaces: Recognition using class specific linear projection,” IEEE
 Transactions on Pattern Analysis and Machine Intelligence, 1997.
[9] T. Ojala, M. Pietikainen, and D. Harwood, “Performance evaluation of texture
measures with classification based on Kullback discrimination of distributions,” in
Pattern Recognition. Vol. 1 - Conference A: Computer Vision amp; Image
Processing., Proceedings of the 12th IAPR International Conference on, 1994.
[10] T. Ojala, M. Pietikainen, and D. Harwood, “A comparative study of ¨ texture
measures with classification based on feature distributions,” Pattern Recognition,
1996.
[11] G. Edwards, C. Taylor, and T. Cootes, “Interpreting face images using active
appearance models,” Proceedings of the Third IEEE International Conference
on Automatic Face and Gesture Recognition, 1998
[12] T. F. Cootes, G. J. Edwards, and C. J. Taylor, Active appearance models., 1998.

[13] M. Turk and A. Pentland, “Eigenfaces for Recognition,” Journal of Cognitive

Neuroscience, vol. 3, no. 1, pp. 71–86, 1991.

[14] W. Haider, H. Bashir, A. Sharif, I. Sharif, and A. Wahab, “A Survey on Face

Detection and Recognition Techniques,” Research Journal of Recent Sciences,
2014.

[15] D. L. Swets, “Using discriminant eigenfeatures for image retrieval,” IEEE

Transactions on Pattern Analysis and Machine Intelligence, vol. 18, no. 8, pp.
831–836, 1996.

[16] L. C. Paul and A. A. Sumam, “Face Recognition Using Principal Component

Analysis Method,” vol. 1, no. 2012, pp. 135–139, 2012.

[17] I. Dagher, “Incremental PCA-LDA algorithm,” in CIMSA 2010 - IEEE

International Conference on Computational Intelligence for Measurement
Systems and Applications, Proceedings, 2010.

[18] E. Rekha and P. Ramaprasad, “An efficient automated attendance management

system based on Eigen Face recognition,”7th International Conference on Cloud
Computing, Data Science & Engineering - Confluence, vol. 5. IEEE, pp. 605–
608, jan 2017.

[19] S. Lukas, A. R. Mitra, R. I. Desanti, and D. Krisnadi, “Student attendance system

in classroom using face recognition technique,” International Conference on
Information and Communication Technology Convergence (ICTC). IEEE, pp.
1032–1035, oct 2016.

[20] N. K. Jayant and S. Borra, “Attendance management system using hybrid face
recognition techniques,”Conference on Advances in Signal Processing (CASP),
vol. 2, no. 11. IEEE, pp. 412–417, jun 2016.

[21] A. Khatun, A. K. M. F. Haque, S. Ahmed, and M. M. Rahman, “Design and

 implementation of iris recognition based attendance management system,”
International Conference on Electrical Engineering and Information
Communication Technology (ICEEICT). IEEE, pp. 1–6, may 2015.

[22] P. Wagh, R. Thakare, J. Chaudhari, and S. Patil, “Attendance system based on

face recognition using eigen face and PCA algorithms,” International Conference
on Green Computing and Internet of Things (ICGCIoT). IEEE, pp. 303–308 ,oct
2015.

[23] Y. Kawaguchi, “Face Recognition-based Lecture Attendance System,” The 3rd

AEARU . . . , no. October, 2005.

[24] N. Kar, M. K. Debbarma, A. Saha, and D. R. Pal, “Study of Implementing

Automated Attendance System Using Face Recognition Technique,”
International Journal of Computer and Communication Engineering, vol. 1, no. 2,
pp. 100–103, 2012.

[25] A. S. S. NAVAZ and T. D. S. P. MAZUMDER, “Face Recognition using Principal

Component Analysis and Neural Networks,” vol. 1, no. April, pp. 91–94, 2001

[26] K. Susheel Kumar, S. Prasad, V. Bhaskar Semwal, and R. C. Tripathi, “Real

Time Face Recognition Using AdaBoost Improved Fast PCA Algorithm,”
International Journal of Artificial Intelligence & Applications, vol. 2, no. 3, pp. 45–
58, jul 2011.

[27] J.-M. Kim and M.-A. Kang, “A Study of Face Recognition Using the PCA and
Error Back-Propagation,” in 2010 Second International Conference on Intelligent
Human-Machine Systems and Cybernetics, vol. 2, no. 1. IEEE, , pp. 241–244,
aug 2010.
[28] R. Toufiq and M. R. Islam, “Face recognition system using PCA-ANN technique
with feature fusion method,” Electrical Engineering and Information &
Communication Technology (ICEEICT), International Conference on, pp. 1–5,
2014.

[29] M. Mazloum and S. Ayat, “Combinational Method for Face Recognition: Wavelet,
PCA and ANN, “Digital Image Computing: Techniques and Applications. IEEE,
pp. 90–95,2008.

[30] R. Chellappa, C. L. Wilson and S. Sirohey, "Human and machine recognition of

faces: a survey," Proceedings of the IEEE,Vol. 83, No. 5, 705-740, May 1995