You are on page 1of 4

International Conference on Intelligent Sustainable Systems (ICISS 2019)

IEEE Xplore Part Number: CFP19M19-ART; ISBN: 978-1-5386-7799-5

Implementation of IOT based Attendance


Management System on Raspberry Pi
1st V Ruhitha 2nd V N Prudhvi Raj 3rd G Geetha
IV Year B.Tech,EIE Department Assoc.Prof, EIE Department Asst.Prof, Dept. of Information Technology
VR Siddhartha Engineering College VR Siddhartha Engineering College VR Siddhartha Engineering College
Vijayawada, India Vijayawada, India Vijayawada, India
ruhitha.v27@gmail.com nagaprudhvi@gmail.com geetaguttikonda@gmail.com

Abstract—The biometric-based attendance monitoring system The best biometric method to take attendance is face
replaces the manual method of taking attendance and reduces recognition [2]. In this method, each student’s face is
the time to take attendance and avoids proxy. There are many collected and stored in the database. When the student is in
biometric methods developed based on the fingerprint recogni-
tion, DNA matching, face recognition, Iris recognition and voice front of the camera it takes images, compares those images
recognition. Out of all, the methods based on face recognition are with the database and marks the attendance. This method
proved to be the best methods. In this paper we are presenting requires less time and provides accurate results, improves
the implementation of attendance monitoring system based on security levels and low fraud rate. This face recognition
the facial recognition. In this method the image database of method can widely be used in various sectors like health care,
the classroom is created first and the database is used to train
the system. Then in recognition phase the camera is fixed at banking, advertising, retail, universities, corporate offices.
the focal point in the classroom and it will capture the images,
then from the captured images student’s faces are recognized &
compared with the database and lastly the attendance is marked II. LITERATURE SURVEY
based on the faces identified. The marked attendance is directly There are many feature extraction and recognition
reported to the students/parents and to the faculty handling the
respective course. In this method, Haar cascade classifier is used algorithms which extract the features from the image and
to detect the faces and captured images are compared with the recognize them with the existed database.
database by using local binary pattern histogram algorithm. The
face recognition system is implemented on the Raspberry Pi 3B Eigen Faces algorithm was developed by Sirovich and
hardware. Kirby (1987) and used by Matthew Turk and Alex Pentland
Index Terms—Automatic attendance,Biometric,Face detection,
Face Recognition,LBPH, Haar Classifier.
in face classification [3]. This algorithm captures the variation
in the collected faces and compares it with individual faces
by generating a set of eigenvectors from the image by the
I. INTRODUCTION
principal component analysis. The eigenvectors are derived
Many universities, colleges, and schools follow the based on the covariance matrix of the probability distribution
conventional face to face attendance marking. This traditional over the high-dimensional vector space of face images.
method requires a lot of time, sometimes students may
not respond to their respective numbers and it involves Fisher faces algorithm is derived by an idea suggested
errors during the manual calculation of attendance. So to by R.A.Fisher in 1936. This algorithm is an enhancement
overcome these problems there is a necessity for an automatic to Eigen faces that uses linear discriminant analysis. The
attendance monitoring system. This automatic attendance linear discriminant analysis finds the linear combination of
monitoring system involves biometric processes. One of features that characterizes or separates two or more classes of
the biometric methods is fingerprint identification [1]. In faces. The resulting combination is used for the reduction of
this method, the fingerprints of each student is collected dimensionality. This algorithm is used when there is a large
and are stored in the database. When the thumb is placed variation in the illumination and facial expressions [4].
on the fingerprint sensor it compares it with the database
and takes attendance. But the disadvantage in this method Local binary pattern histogram algorithm is described in the
is the students should wait in the queue to give their year 1994 and was a powerful technique for feature extraction.
prints so it is a time taking process. The other method of In this algorithm, the grayscale image is divided into cells of
biometric is iris. In this method, the eyeball or iris of each the 3X3 matrix. The obtained matrix is represented with the
student is taken and stored in the database. when eyeballs range of pixel intensities. The center value of the matrix is
matches with the database it takes attendance. This is a time- considered as the threshold and each pixel value is compared
consuming method and in practice, it is difficult to implement. with the threshold. If the pixel value is greater than the
threshold it is given 1 otherwise 0. An 8-digit binary value

978-1-5386-7799-5/19/$31.00 ©2019 IEEE 584


Authorized licensed use limited to: Siksha O Anusandhan University. Downloaded on November 14,2023 at 07:52:36 UTC from IEEE Xplore. Restrictions apply.
International Conference on Intelligent Sustainable Systems (ICISS 2019)
IEEE Xplore Part Number: CFP19M19-ART; ISBN: 978-1-5386-7799-5

and is converted into decimal value for convenience. Then A. Capturing images
the histogram is computed over the cells and frequency of The camera module is placed at the focal point in the
each number and concatenate the histogram of all cells to classroom or office. A camera module is used for capturing
obtain the feature vector. This feature vector is processed to the samples of the classroom. These samples are used for
classify the images [2], [5]. recognition of faces and to mark the attendance.

Scale-invariant feature transform algorithm is published by B. Creating database


David Lowe in 1999. This algorithm detects and describes As biometric method has been chosen for marking the
the local features in the images. Whenever the image is attendance it is important for enrolment of every individual
transformed, like the size of the image is altered or when whose attendance has to be taken. The face of every individual
it is rotated the features of the image is invariable. These is captured, converted into grayscale and are stored in the
invariable features are used to detect the image [6]. database which includes a person’s identity number. Here
multiple samples are taken for each individual with different
Speed up robust feature algorithm was presented by Herbert facial expressions.
Bay, Tuytelaars, T. and Van Gool in the year 2006. This
algorithm is patented local feature detector and descriptor.
SURF descriptors are used to locate and recognize objects,
people or faces, to reconstruct 3D scenes, to track objects
and to extract points of interest. This algorithm uses a multi-
resolution pyramid technique where the image is transformed
into coordinates to copy the original image with Pyramidal
Gaussian shape to obtain an image with the same size but
with decreased bandwidth [7] .

Many advanced algorithms are developed in the last decade.


But these algorithms demand more hardware requirements
than the present hardware used to execute them without any
Fig. 2. Database of the captured faces
difficulty [8]–[14].

C. Detecting faces
III. I MPLEMENTED S YSTEM The faces in the image are detected by Haar cascade
classifier. This method was method proposed by Paul Viola and
The implemented system can be used for taking the Michael Jones. Haar cascade classifier is available in OpenCV
attendance in universities, colleges, and offices. The source library which contains pre-trained classifiers for face,
architecture of the system is shown in Figure 1. A webcam eye, smile, etc. Haar cascade classifier mainly classifies the
face based on the edge features and corner features. This
classifier has high detection accuracy and low false positive
rate.
D. Preprocessing
The captured image may contain unnecessary background
noise and elements other than faces. It is important to remove
such elements. Feature extraction is the method that reduces
the image to only face available in the image. By this method,
the image is reduced to a size of 150x150.
E. Face recognition
Face recognition involves two steps of training and feature
Fig. 1. Block diagram of the system extraction. To recognize faces we should train the algorithm
with images and their respective identity numbers. Feature ex-
is connected to Raspberry pi3 b+ and it is placed at the focal traction is done by many face recognition algorithms available
point in the classroom. This webcam captures the images in OpenCV like eigenfaces, Fisher faces, local binary pattern
and extracts human faces. Then recognize the faces with the histogram(LBPH).LBPH algorithm is simple and widely used
existing database through library files present in the openCV. to recognize faces. In this algorithm, each image is analyzed
The steps involved are as follows. independently. In this algorithm, the grayscale image is di-
vided into cells of the 3X3 matrix. The obtained matrix is

978-1-5386-7799-5/19/$31.00 ©2019 IEEE 585


Authorized licensed use limited to: Siksha O Anusandhan University. Downloaded on November 14,2023 at 07:52:36 UTC from IEEE Xplore. Restrictions apply.
International Conference on Intelligent Sustainable Systems (ICISS 2019)
IEEE Xplore Part Number: CFP19M19-ART; ISBN: 978-1-5386-7799-5

represented with the range of pixel intensities. The center


value of the matrix is considered as the threshold and each
pixel value is compared with the threshold. If the pixel value
is greater than the threshold it is given 1 otherwise 0. An
8-digit binary value and is converted into decimal value for
convenience.Then histogram is computed over the cells and
frequency of each numbers and concatenate the histogram of
all cells to obtain the feature vector. This feature vector is
processed to classify the images [2], [5], [15]. An image in
Local binary pattern histogram is mathematically represented
as,
p−1
X
LBP (xn , yn ) = 2p S (ip − ic ) (1)
p=0

ip is the central pixel intensity and ic is the intensity of the


neighbouring pixel. S is the signum function defined as

1 x≥0
S (x) = (2)
0 else

This description enables to capture very fine grained details


in images. This to align an arbitrary number of neighbors on
a circle with a variable radius, which enables to capture the
neighborhoods shown in figure 3.

Fig. 4. Flowchart of the algorithm

A. Hardware Software Used


Fig. 3. Neighbourhood with variable radius to extract features Raspberry Pi 3b:Broadcom BCM2837B0, Cortex-
A53 (ARMv8) 64-bit SoC @ 1.4GHz, 1GB LPDDR2
SDRAM, 2.4GHz and 5GHz IEEE 802.11.b/g/n/ac wireless
For a given Point (xn , yn ) the position of the neighbour
LAN, Bluetooth 4.2, BLE, Gigabit Ethernet over USB 2.0
(xp , yp ) can be calculated by using the following equation
(maximum throughput 300 Mbps), 4 x USB 2.0 ports, HDMI,

2Πp
 CSI camera port for connecting a Raspberry Pi camera,
xp = xc + R cos (3) DSI display port for connecting a Raspberry Pi touchscreen
P
display, 2.0 graphics, OpenGL ES 1.1, MPEG-4 decode,
where R is the radius of the circle and P is the number of H.264, H.264 Encode. Micro SD port for loading your
sample points. operating system and storing data.

IV. EXPERIMENTAL SETUP AND RESULTS Software:Raspbian Operating System, OpenCV Image
Processing Library
In this work a collection of 5 students along with 50 (https://opencv.org), AWS IOT Sdk.
images of each individual person are taken and are stored
in the database with their respective identity number. These In this developed system if the occurrence of the face in
50 images are trained to recognize faces. The camera module captured images is greater than 50 then attendance is marked
which is placed at the focal point in the classroom captures as present otherwise absent. As a result, excel sheet with
the 100 images of the classroom and recognize the faces in report of attendance is created, the list of absentees is mailed
the captured images. If the probability of face is greater than to the head of department and sms is sent to the parents of
50 then it is marked as present otherwise absent. As a result, absentees. The same data will be pushed into the Amazon
excel sheet with report of attendance is created, the list of Web Services cloud using AWS IOT sdk and the same can be
absentees is mailed to the head of department and SMS is accesed through the android app from anywhere in the world.
sent to the parents of absentees.

978-1-5386-7799-5/19/$31.00 ©2019 IEEE 586


Authorized licensed use limited to: Siksha O Anusandhan University. Downloaded on November 14,2023 at 07:52:36 UTC from IEEE Xplore. Restrictions apply.
International Conference on Intelligent Sustainable Systems (ICISS 2019)
IEEE Xplore Part Number: CFP19M19-ART; ISBN: 978-1-5386-7799-5

R EFERENCES
[1] M. J. Cheok, Z. Omar, and M. H. Jaward, “A review of hand gesture and
sign language recognition techniques,” International Journal of Machine
Learning and Cybernetics, vol. 10, no. 1, pp. 131–153, 2019.
[2] T. Ahonen, A. Hadid, and M. Pietikainen, “Face description with local
binary patterns: Application to face recognition,” IEEE Transactions on
Pattern Analysis and Machine Intelligence, vol. 28, pp. 2037–2041, Dec
2006.
[3] M. Turk and A. Pentland, “Eigenfaces for recognition,” Journal of
cognitive neuroscience, vol. 3, no. 1, pp. 71–86, 1991.
[4] P. N. Belhumeur, J. P. Hespanha, and D. J. Kriegman, “Eigenfaces vs.
fisherfaces: Recognition using class specific linear projection,” tech. rep.,
Yale University New Haven United States, 1997.
[5] T. Ojala, M. Pietikainen, and T. Maenpaa, “Multiresolution gray-scale
and rotation invariant texture classification with local binary patterns,”
Fig. 5. Face Detection and feature extraction phase IEEE Transactions on pattern analysis and machine intelligence, vol. 24,
no. 7, pp. 971–987, 2002.
[6] D. G. Lowe, “Distinctive image features from scale-invariant keypoints,”
International journal of computer vision, vol. 60, no. 2, pp. 91–110,
2004.
[7] H. Bay, A. Ess, T. Tuytelaars, and L. Van Gool, “Speeded-up robust
features (surf),” Computer vision and image understanding, vol. 110,
no. 3, pp. 346–359, 2008.
[8] Z. Sun, Z.-p. Hu, M. Wang, and S.-h. Zhao, “Dictionary learning
feature space via sparse representation classification for facial expression
recognition,” Artificial Intelligence Review, vol. 51, no. 1, pp. 1–18,
2019.
[9] X. Song, Y. Chen, Z.-H. Feng, G. Hu, T. Zhang, and X.-J. Wu,
“Collaborative representation based face classification exploiting block
weighted lbp and analysis dictionary learning,” Pattern Recognition,
vol. 88, pp. 127–138, 2019.
[10] Y. Makhija and R. S. Sharma, “Face recognition: Novel comparison of
various feature extraction techniques,” in Harmony Search and Nature
Inspired Optimization Algorithms, pp. 1189–1198, Springer, 2019.
[11] I. M. Sayem and M. S. Chowdhury, “Integrating face recognition security
system with the internet of things,” in 2018 International Conference on
Fig. 6. Face recogntion and attendance recording phase Machine Learning and Data Engineering (iCMLDE), pp. 14–18, Dec
2018.
[12] J. Lu, X. Fu, and T. Zhang, “A smart system for face detection with
spatial correlation improvement in iot environment,” in 2017 IEEE
SmartWorld, Ubiquitous Intelligence Computing, Advanced Trusted
Computed, Scalable Computing Communications, Cloud Big Data
Computing, Internet of People and Smart City Innovation (Smart-
World/SCALCOM/UIC/ATC/CBDCom/IOP/SCI), pp. 1–4, Aug 2017.
[13] I. Aydin and N. A. Othman, “A new iot combined face detection of
people by using computer vision for security application,” in 2017
International Artificial Intelligence and Data Processing Symposium
(IDAP), pp. 1–6, Sep. 2017.
[14] Y. Chen, Q. Chen, K. Chou, and R. Wu, “Low-cost face recognition
system based on extended local binary pattern,” in 2016 International
Automatic Control Conference (CACS), pp. 13–18, Nov 2016.
[15] P. Viola and M. Jones, “Rapid object detection using a boosted cascade
of simple features,” in Computer Vision and Pattern Recognition, 2001.
Fig. 7. Attendance recorded to excel sheet CVPR 2001. Proceedings of the 2001 IEEE Computer Society Confer-
ence on, vol. 1, pp. I–I, IEEE, 2001.

V. CONCLUSION

The implemented attendance monitoring system is per-


forming with 96 percent accuracy. The system is using haar
classifier and LBPH algorithms for face detection and face
recognition. There are many algorithms develeloped in the
recent years based on wavelet transforms, support vector
machines, neural networks and deeplearning techniques. But
these advanced techniques requires the most sophisticated
hardware such as high end CPU’s and GPU’s. The cost of
the system is also increases with the advanced hardware. So
the implemented system is economical and very useful to
implement with low cost.

978-1-5386-7799-5/19/$31.00 ©2019 IEEE 587


Authorized licensed use limited to: Siksha O Anusandhan University. Downloaded on November 14,2023 at 07:52:36 UTC from IEEE Xplore. Restrictions apply.

You might also like