Intrusion Detection System Using Face Recognition

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 10 Issue: 03 | Mar 2023 www.irjet.net p-ISSN: 2395-0072
© 2023, IRJET | Impact Factor value: 8.226 | ISO 9001:2008 Certified Journal | Page 1222
Intrusion Detection System Using Face Recognition
Ayush Dubey1, Ajay Pandey2, Riya Pandey3, Zuhair Bhati4, Reena Kothari5
1,2,3,4,5 Department of information Technology, Shree LR Tiwari College of Engineering, Maharashtra, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - The Intrusion Detection System Using Face
Recognition is a security system that uses facial recognition
technology to detect and prevent unauthorized access. This
system works by capturing images of people attempting to
gain access to a secure area and comparing them to a
database of authorized personnel. The systemcanidentify and
alert security personnel if there is a match with an
unauthorized person. The proposed system utilizes machine
learning algorithms and deep learning techniques for
improved accuracy and reliability. This system uses a
Raspberry Pi, Sensor, Camera, Email services and Python &
Shell Script. The system can be used in various settings,
including airports, banks, and government institutions, to
enhance security and prevent potential security breaches.
Key Words: Security System, Facial Recognition, Raspberry
Pi, Python Scripts, Machine Learning.
1.INTRODUCTION
Security has always been a top priority for organizations
seeking to protect their assets and prevent unauthorized
access. Traditional security measures such as locks, badges,
and passwords have been effective to some extent, but they
are not foolproof. As technology continues to evolve,
organizations need to adopt advanced security measures
that offer higher accuracy and reliability. One such measure
is the Intrusion Detection System Using Face Recognition.
This is an advanced security system that utilizes facial
recognition technology to detect and prevent unauthorized
access. This system captures images of individuals
attempting to gain access to restricted areas and compares
them to a database of authorized personnel. If there is a
match with an unauthorized person, the system alerts
security personnel, who can then take appropriate action.
This paper aims to provide a comprehensive understanding
of the Intrusion Detection System Using Face Recognition
and its potential to enhance security measures in
organizations.
The objective is to offer insights into the working principles
of the system, its advantages, its applications, and its
limitations, providing a foundation for future research and
development.
2. PROPOSED SYSTEM
Fig -1: Flowchart
The proposed system for the Intrusion Detection System
Using Face Recognition is designed to provide a highly
reliable and efficient way to detectandpreventunauthorized
access. The system comprises the following components:
It uses Hardware such as
• Raspberry Pi
• Camera Module
• Jumper Cables
• Motion Sensor

And software such as
• Python Scripts
• Mutt email client
• Raspberry Pi-Camera Interfacing
• GPIO Pins Programming
This system uses Face Detection component which is
responsible for detecting and localizing faces in the input
image. It utilizes algorithms such as Haar cascades or deep
learning techniques such as Convolutional Neural Networks
(CNNs) to accurately detect and extract faces from the input
image.
The main component of this project is the Raspberry Pi, a
small computer that powerstheintrusiondetectionsystem's
backend operations. We use Python and shell script to code
all of our programs.
Functional setup of IDS components:
Fig -2: Proposed System containing all the functional units
Due to its user-friendly features and cost advantages,
Raspberry Pi has been chosen as the system's processing
unit. The Raspberry Pi has been given the algorithmic rule
that was programmed in Python. It is programmed so that it
first sets all of the GPIO Pins to provide a correct interface
between all of the attached devices and the Raspberry Pi,
after which all of the connected modules will carry out their
functions as intended.
3. Algorithm
Below is a list of algorithms analyzed for the Facial
Recognition Process:
3.1 CNN
Convolutional Neural Networks (CNNs) are a type of neural
network that are commonly used for image classification,
object detection, and other computer vision tasks. They are
made up of a series of layers, including convolutional layers,
pooling layers, and fully connected layers, that allow the
network to learn hierarchical representations of the input
data. This makes CNNs particularlyeffectivefortaskssuchas
object recognition, where the network needs to be able to
identify objects at different scales and orientations.
3.2 SVM
The support vector machine (SVM) is a classification
technique applied on linear as well as nonlinear data. It is a
composite version of KNN combined with SVM for image
catalog recognition and is increased in. [7] In this algorithm,
training is done with the help of the nearest K the neighbors
of the data point are not labeled. First, K- nearest data points
are determined. Then pair the distance betweentheseKdata
points is calculated. Hence, we get a distancematrixfromthe
calculation distance. The Kernel matrix is then designed
from distance matrix is obtained. This core matrix is
provided as input to the SVM classifier. Theresultistheclass
of the data point is unknown. In addition, a can use SVM but
time consuming is one of the defects. It also involves
calculation pair distances.
3.3 FisherFaces
FisherFaces is a technique for face recognition that uses
linear discriminant analysis (LDA) to finda low-dimensional
representation of face images that maximizestheseparation
between different individuals. The basic idea behind
FisherFaces is to project high-dimensional face images onto
a lower-dimensional subspace where each dimension
corresponds to a linear combination of pixel values. This
projection is done in a way that maximizes the ratio of
between-class variance to within-class variance, which
effectively separates the face images corresponding to
different individuals. To perform face recognition using
FisherFaces, a set of training images is first used to learn the
subspace projection. This involves computing the meanface
and eigenvectors of the covariance matrixofthefaceimages,
and selecting the top eigenvectors that maximize the Fisher
criterion. Once the subspace projectionislearned,a newface
image can be projected onto the subspace, and compared to
the training images using a nearest neighbor or other
classification algorithm.
4. Literature Review
Mehek Male, Sahil Colvalkar, Ajay Pandey, and Sarvesh
Pandey, [1] developed a security system that uses motion
detection to identify potentialintrudersinaprotectedarea.It
is designed to be used in a variety of applications, including
home security, commercial security, and military and law
enforcement operations. The system works by using motion
sensors to detect any movements within a protected area.
When motion is detected, the system sends an alert to the
user or security personnel. The system can be configured to

work with a variety of sensors, including infrared sensors,
ultrasonic sensors, and microwave sensors.Itappearstobea
promising solution for detecting intruders and enhancing
security in a variety of settings. However, its effectiveness
and suitability may depend on various factors such as the
specific application, environment, and the quality of sensors
and system setup. It is always recommended to evaluate the
system's capabilities and limitations before implementing it
in a particular setting.
Usman Shuaibu Musa, Megha Chhabra and Aniso Ali,
Mandeep Kaur, [2] published their research work on
Intrusion Detection System (IDS) using Machine Learning
(ML) techniques which has become an increasingly popular
research area in recent years. IDS is a security mechanism
that identifies malicious activities and potential security
breaches in computer systems. ML techniques provide a
promising approachtoimprovetheaccuracyandefficiencyof
IDS.
Irving Vitra Paputungan, Mahbub Ramadhan Al Fitri and
Unan Yusmaniar Oktiawati [3] presented a research paper
which explores the design and development of a DIY home
security system that utilizesmotionandmovementdetection
technology toenhancesecuritymeasures.Thesystemusesan
infrared sensor to detect movement within a specific range
and triggers an alarm if the movement is detected within the
user-defined boundary.
Lixiang Li, Xiaohui Mu, Siying Li and Haipeng Peng [4] have
proposed a comprehensive and insightful overview of the
current state of face recognition technology. The article
provides a thorough introduction to the history and
development of face recognition technology, as well as a
detailedexplanationofthevarioustechniquesandalgorithms
used in modern facial recognition systems. The authors also
explore the ethical and social implications of the technology,
including privacy concerns and potential biases.Overall, this
review provides a valuable resource for anyoneinterestedin
understanding the technical and societal implications of face
recognition technology. It is well-researched,clearlywritten,
and provides a balanced perspective on the topic.
M. Khan, S. Chakraborty, R. Astya and S. Khepra [5]published
a Face detection and picture or video recognition which is a
popular subjectofresearchonbiometrics.Facerecognitionin
a real-timesetting has an exciting areaand a rapidly growing
challenge. Framework for the use of face recognition
application authentication. This proposes the PCA (Principal
Component Analysis) facial recognition system.
5. RESULTS & ANALYSIS
Let us just go through the components used in this Intrusion
detection system and then analyze the output results.
Fig -3: Passive Infrared Sensor
Fig -4: Pin layout of the PIR (Passive Infrared) sensor
Fig -5: Raspberry Pi 3 Model B v1.2 (Front)
Fig -6: Raspberry Pi 3 Model B v1.2 (Back)

Fig -7: Pi-Camera
Fig -8: Connection of the Pi-Camera to the Raspberry Pi
Fig -9: Output Screenshots from Pi-Camera
6. CONCLUSIONS
In today’s generation, everything relies on computationand
data either directly or indirectly. This project has hardware
and software package. Hardware describes how the system
was designed, what module will it use. The system is meant
for un-welcomed person detection. An Intrusion Detection
System using Face Recognition is a sophisticated security
system that uses advanced algorithms and machinelearning
techniques to detect and prevent unauthorized access to a
secure area. It works by capturing images of people who are
attempting to enter a restricted area, and then comparing
those images against a database of known faces. If the
system detects a match, it grants access; otherwise, it denies
entry and raises an alarm. One of the key advantages of this
system is that it is highly accurate and can operate in low-
light conditions, making it suitable for use in a variety of
environments. It is also a non-intrusive method of access
control, which can be important in situations where privacy
is a concern. However, there are also some potential
drawbacks to using a face recognition-based intrusion
detection system. One concern is that the technology may
not be completely reliable, especially if the lighting
conditions are poor or if the person being scanned is
wearing a disguise or mask. Additionally, there arepotential
privacy concerns associated with using facial recognition
technology, particularly if the system is storing and
analyzing biometric data. The projected system provides
digital computer primarily based home security system by
use of terribly advanced low price stable software package.
7. FUTURE SCOPE
By adding an alternative energy panel,thecamera isgoingto
be capable of gathering the solar energy and be wireless. We
can add a feature, such as if intrusion is detected then
automatically all doors will be locked, so the intruder wont

escape. We can send emergency signal to crime department.
We can add alarm systems which will alert the security
guards. There is always room for improvement when it
comes to the accuracy of face recognition algorithms. In the
future, we can expect to see more sophisticated techniques
that can better handle factors such as changes in lighting,
pose, and expression. IDS can be integrated with other
technologies, such as artificial intelligence and machine
learning, to enhance its performance. For example, IDS can
learn from past security breaches and automatically adjust
its rules and algorithms to detect new threats.
REFERENCES
[1] Mehek Male, Sahil Colvalkar, Ajay Pandey, and Sarvesh
Pandey, “S.W.A.T – Motion Based Intrusion Detection
System,” (International Research Journal ofEngineering
and Technology (IRJET) 01 | Jan-2018) p-ISSN: 2395-
0072; e-ISSN: 2395-0056.
[2] Usman Shuaibu Musa, Megha Chhabra and Aniso Ali,
Mandeep Kaur, ”Intrusion Detection System using
Machine Learning Techniques”, (Published in: 2020
International Conference on Smart Electronics and
Communication (ICOSEC) Date Added to IEEE: 07
October 2020)
[3] Irving Vitra Paputungan, MahbubRamadhanAl Fitriand
Unan Yusmaniar Oktiawati, “Motion and Movement
Detection for DIY Home Security System”,(Publishedin:
2019 IEEE Conference on Sustainable Utilization and
Development in Engineering and Technologies
(CSUDET)),Penang,Malaysia,2019,pp.122125,doi:10.110
9/CSUDET47057.2019.9214684.
[4] L. Li, X. Mu, S. Li and H. Peng, "A Review of Face
Recognition Technology," in IEEE Access, vol. 8, pp.
139110-139120,2020,doi:
10.1109/ACCESS.2020.3011028.
[5] M. Khan, S. Chakraborty, R. Astya and S. Khepra, "Face
Detection and Recognition Using OpenCV," 2019
International Conference on Computing,
Communication, and Intelligent Systems (ICCCIS),
Greater Noida, India, 2019, pp. 116-119, doi:
10.1109/ICCCIS48478.2019.8974493.

Intrusion Detection System Using Face Recognition

More Related Content

Similar to Intrusion Detection System Using Face Recognition (20)

More from IRJET Journal (20)

Recently uploaded (20)

Intrusion Detection System Using Face Recognition