2016 IEEE International WIE Conference on Electrical and Computer Engineering (WIECON-ECE)

19-21 December 2016, AISSMS, Pune, India

Face Detection and Recognition using Raspberry Pi

Ishita Gupta1, Varsha Patil2, Chaitali Kadam3, Shreya Dumbre4

[email protected]
AISSMS IOIT, Kennedy Road, Pune, India2 Savitribai Phule Pune University, India1, 3, 4

Abstract— in today’s world, face recognition is an important displayed on LCD and the motor will rotate indicating
part for the purpose of security and surveillance. Hence opening and closing of the gate.
there is a need for an efficient and cost effective system. Our
goal is to explore the feasibility of implementing Raspberry
Pi based face recognition system using conventional face
detection and recognition techniques such as Haar detection
and PCA. This paper aims at taking face recognition to a
level in which the system can replace the use of passwords
and RF I-Cards for access to high security systems and
buildings. With the use of the Raspberry Pi kit, we aim at
making the system cost effective and easy to use, with high
performance. Fig 2. Face Bunch Graph

Keywords- Face recognition, face detection, Haar Detection, 2. FACE DETECTION USING HAAR
PCA, Raspberry Pi.
CASCADES
1. INTRODUCTION Object Detection using Haar feature-based cascade
The information age is quickly revolutionizing the classifiers is an effective method which was proposed by
way transactions are completed. There is a need for a Paul Viola and Michael Jones [2, 3]. It is an adaptive
faster and accurate user identification and authentication machine learning based approach in which a cascade
method. Face recognition has become one of the most function is trained from several positive and negative
important user identification methods. Literature survey images. This is then used to detect objects in other
statistics shows that research work in face recognition images. Initially, this algorithm requires plenty of positive
system is in its booming era, and in the past forty years, images (images of faces) and negative images (images
the research in this field has increased exponentially [1]. without faces) to train the classifier. Then Features are
extracted from it.
Every feature is a single value obtained by subtracting
sum of pixels under white rectangle from sum of pixels
under black rectangle. Then all possible sizes and
locations of every kernel are used to calculate plenty of
features. To calculate each feature, the sum of pixels
under white and black rectangles is found [2, 3].
3. FACE RECOGNITION USING EIGEN FACE
APPROACH
The face recognition algorithms used here is
Principal Component Analysis (PCA). It involves a
mathematical procedure that transforms a number of
possibly correlated variables into a number of
uncorrelated variables called principal components,
Fig1. Research Statistics of Face Recognition related to the original variables by an orthogonal
transformation [3]. The Eigen face approach helps
Face recognition technology emulates the capabilities reducing the size of the database required for recognition
of human eyes to detect faces. This is done by smart of a test image. The Eigen values calculated from the
computing that creates “face bunch” that consists of 70 Eigen Vector covariance matrix are rejected or stored
nodal points. Features are extracted from the face and depending upon the threshold thus creating a face space
saved as templates. These templates are compared to the [3, 4, and 5]. Calculating the weights and the Euclidean
face detected. For this research, we interfaced an LCD, distance a comparison is held and match is found [5]. This
Camera and a Motor to the Raspberry Pi board. We have conventional Eigen face Approach is incorporated in the
made a real time application, which compares the scans to ARM Cortex of Raspberry Pi for face recognition using
records stored in the Raspberry Pi which in turn is used as face recognition modules in python code.
a gate pass, wherein the name of the detected person is

978-1-5090-3745-2/16/$31.00 ©2016 IEEE 83

Authorized licensed use limited to: Parul Institute of Engineering and Technology. Downloaded on September 16,2022 at 10:10:32 UTC from IEEE Xplore. Restrictions apply.
 4. METHODOLOGY those stored in database. On the other hand the output part
The Functionality of this system is mainly consist of LCD, DC motor followed by its driver
categorized in following steps. L293D.The motor used here has a range up to 10,000
RPM.
a. To enrol and detect faces using camera
connected to the ARM Cortex of Raspberry Pi board. The necessary steps to initialize the system include
b. To display the match status on the LCD as well creating a database and then using that database as a
as the terminal running on the VGA (Video graphics reference for comparing the live test points. As we boot
array) monitor. the raspberry pi following things needs to be taken into
c. To program for the same using python language. consideration like connecting the LCD monitor in order to
The code imports certain modules that enable functions verify the image captured by the webcam which will be
such as face recognition, GPIO modules. used for detecting the entity.
d. To drive the motor in clockwise and Linux based commands are provided as an input to
anticlockwise direction upon getting a present status. the Raspberry Pi2 in order to save images in the external
SD card. Two separate directories are maintained for the
The identification and authentication technology operate image which are captured as a reference i.e. Positive and
using the following four stages: [6] Negative, Positive directory holds the reference image
which has been captured with no error. While those with
Capture Extraction Comparison minimal to maximum defective images are stored in the
negative directory.
5. SYSTEM DESIGN
Hardware Design:
Accept/ Match/No-
Reject Match It includes Raspberry Pi2development kit, connecting
cables, LCD, DC Motor driver IC, DC motor, Power
Fig3. Flow of operation Supply, USB keyboard, USB mouse and USB webcam.

a. Capture: A physical or behavioural sample is 1. Raspberry Pi 2

captured by the system during enrolment and also in Raspberry Pi is a small scale based computer, it
identification or verification process performs efficiently when it comes to running games,
b. Extraction: Unique data is extracted from the image files, and documents. They can be manufactured
sample and a template is created. using different configurations as per the user’s needs.

c. Comparison: The template is then compared with an It runs at 700MHZ with nearly 256MB RAM, with on
existing sample. board graphic card capabilities. RaspberryPi2 are
considered one of the few low power consumption CPU’s
d. Match/non match: The system decides if the [7].
features extracted from the new samples are a match or a
non match and accordingly accept/reject. The basic layout of the Raspberry pi2 consist of
1. Input/Output.
2. RAM
3. CPU/GPU
4. USB hub
5. Ethernet
6. HDMI Port

2. LCD
Fig4. Simplified Block Diagram The LCD used here is 16 x 2 alphanumeric LCD. It is a
very basic module which overcomes the disadvantage of
the seven segment display as well as multi-segment LCD.
The input/output modules are interfaced to the It consists of command register to perform the user
Raspberry Pi 2. The input part, the keyboard and mouse defined instructions and data register to display the data
are used to pass commands to the Raspberry pi and on LCD.
Webcam to capture image which are stored in Secure
Digital Card and is also used to compare an image from

84

Authorized licensed use limited to: Parul Institute of Engineering and Technology. Downloaded on September 16,2022 at 10:10:32 UTC from IEEE Xplore. Restrictions apply.
 3. DC Motor Driver
The L293D Motor drivers provide voltages ranging from
5V TO 36 V with output current of 1 A per channel.
L293D provides bi directional currents and can be
operated at temperature up to 70 degree Celsius.
4. DC Motor
The speed of the motor can be changed by varying the
voltage levels. Here the motor rotates in clockwise and
anticlockwise direction at a speed of 10,000RPM,
representing opening and closing of gate.

Software design:
To code for the Hardware setup, python language is used
[8]. To access the editor Linux terminal is required. The
face detection and recognition part is carried out by
certain Linux commands which are as follows:

• sudo su # To switch to root user.

• python capture-positive.py # To create
database of images.
• ctrl+c # To give keyboard interrupt and stop the
image capturing after desired number of images
are taken for a face.
• python train.py # To train the captured images
and extract features from it, by separating the
face (positive images) from the background or
non-face image (negative images) and thus
creating a .xml file.
• python project2.py # To detect and recognise
which face is present in front of the web cam. It
calculates the confidence value and matches the
flag status. If a face corresponds to certain flag
10 times then it recognises the respective face
and indicates a present status on the terminal as
well as the LCD and the motor rotates in the
clockwise direction for 15 seconds and
anticlockwise direction for 5 seconds, indicating
opening and closing of gate

Fig5. Flow Chart of Software functioning

85

Authorized licensed use limited to: Parul Institute of Engineering and Technology. Downloaded on September 16,2022 at 10:10:32 UTC from IEEE Xplore. Restrictions apply.
 6. OUTPUT RESULT
To recognise correct face it must match at least 10 times
with an existing flag.

Fig8. Output on LCD

7. ACKNOWLEDGEMENT
We would like to express our sincere gratitude to the staff
of Department of Electronics and Telecommunication,
AISSMS IOIT, Pune for their encouragement and
Fig6. Identification of Face support. We would also like to thank the faculty of ARDE
(DRDO), Pune who helped us in this research work.
When face images are captured and trained several
positive and negative images are created. Capture.pgm, 8. REFERENCES
positive.png, negative.png and mean.png files are created [1] W. Zhao, R. Chellappa, P. J. Phillips, A. Rosenfeld, "Face
in the first instance after obtaining the confidence value. Recognition: A Literature Survey", ACM ComputingSurveys,
Vol. 35, No. 4, December 2003, pp. 399–458.
[2] R. Lienhart and J. Maydt, "An extended set of Haar-like
features for rapid object detection," Proceedings. International
Conference on Image Processing, 2002, pp. I-900-I-903 vol.1.

[3] P. Viola and M. Jones, "Rapid object detection using a boosted

cascade of simple features," Proceedings of the 2001 IEEE
Computer Society Conference on Computer Vision and Pattern
Recognition. CVPR 2001, 2001, pp. I-511-I-518 vol.
[4] A. M. Patil, Dr. Satish R. Kolhe, Dr. Pradeep M. Patil (2009),
“Face Recognition by PCA Technique”, Second International
Conference on Emerging Trends in Engineering and
Technology, ICETET-09
a. Negative Image b. Positive Image [5] M. A. Turk and A. P. Pentland, "Face recognition using
eigenfaces" Proceedings. 1991 IEEE Computer Society
Conference on Computer Vision and Pattern Recognition,
Maui, HI, 1991, pp. 586-591

[6] U. Gupta, P. Sinha and P. Chharia, "Reduction of genuine and

imposter score overlapping based on intra-class variations and
deviation of scores in a multimodal biometric system," 2015
International Conference on Signal Processing and
Communication Engineering Systems, Guntur, 2015, pp. 126-
131.

[7] Ningthoujam Sunita Devi and K Hemachandran, “Automatic

Face Recognition System using Pattern Recognition
Techniques: A Survey” Proceedings International Journal of
a. Mean Image d. Capture.pgm Computer Applications, IJCA-13

Fig7. Various images created [8] Pritish Sachdeva and Shrutik Katchii, “A Review paper on
Raspberry Pi” Proceedings International Journal of Current
If the face is recognised correctly then the LCD glows and Engineering and Technology, IJCET-14
reads the name of the person detected and a “Present”
status. Similarly at the terminal on the Personal computer [9] Y. C. Huei, "Benefits and introduction to python programming
for freshmore students using inexpensive robots," 2014 IEEE
the name and a “presentOk” status are obtained. International Conference on Teaching, Assessment and
Learning for Engineering (TALE), Wellington, 2014, pp. 12-17

86

Authorized licensed use limited to: Parul Institute of Engineering and Technology. Downloaded on September 16,2022 at 10:10:32 UTC from IEEE Xplore. Restrictions apply.