International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 10 Issue: 08 | Aug 2023 www.irjet.net p-ISSN: 2395-0072

Face Recognition System using OpenCV

Arun Binoy1, Acsahmol Edwin 2, Jayakrishnan K3 , Pranav S4

1-4 BTECH UG Students, Department of Computer Science and Engineering, TOMS college of Engineering
APJ Abdul Kalam Technological University, Kerala, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - The face recognition system using OpenCV is a 1.1 SCOPE
system which recognizes know faces and store it with time
stamp. Then recognize the unknown faces and store it in The suggested system aims to create and deploy a
another dataset with time stamp. The versatile face reliable and effective system for facial recognition
recognition system utilizing OpenCV, designed to streamline technology-based automated attendance tracking. In order to
attendance tracking in online classes, meetings, and effectively detect and identify people in photos or videos, the
residential security applications. Leveraging advanced image project will use facial recognition algorithms and techniques,
processing techniques, the system identifies and verifies which will replace the need for manual attendance
participants' faces in real-time. Its adaptable nature allows procedures. The scope most likely includes activities like
easy customization of recognized individuals, facilitating image preprocessing and enhancement, face detection and
efficient management of attendance lists. This technology recognition techniques implementation, facial feature
proves beneficial for remote learning scenarios, virtual extraction, and creation and management of a face database.
meetings, and enhancing security protocols in residential Creating an intuitive user interface, connecting the system
settings. By seamlessly integrating with existing platforms, the with current attendance management systems, and assessing
system offers a user-friendly and efficient solution to automate the system's performance in terms of accuracy, speed,
attendance management, thereby improving overall scalability, and usability are all potential components of the
convenience and security in various contexts. project.

Key Words: face recognition, OpenCV, dataset, time stamp, 1.2 FACE DETECTION
real time recognition.
Changes in an object's position in respect to its
1.INTRODUCTION surroundings are monitored using face detection. One of the
most important security elements in recent years is face
Face recognition is important because it not only allows us to detection software. It is used to enhance security equipment
use our faces as keys, but it also allows face recognition that has already been installed, such as the motion sensor
systems to read our expressions in real time. Face illumination on indoor and outdoor security cameras. An
recognition is rapidly improving as the Internet of Things advanced facial identification system like this might be
grows and new gadgets are developed. Everyone's first automated to detect criminals by using CCTV cameras
concern in the current world is security. People are harassed positioned at numerous locations. The goal of the project is
at home, and the antiquated security mechanisms designed to create an automated system that effectively and reliably
to keep people safe have failed. Various electronic devices, tracks attendance using facial recognition technology.
such as mobile phones, laptops, and ATMs, use biometric
authentication or passcodes, however these can be easily
accessed by thieves through any methods, making them
insecure. Every face is unique because it can be recognised,
which is essential for determining a person's identity. Facial
recognition is a novel biometric technology for criminal
identification that offers high accuracy with low intrusion. It
is a technique that uses facial recognition to automatically
identify and validate people in video or image frames. This
study describes a face recognition system that incorporates
the best face detection, feature extraction, and classification
approaches currently in use. MTCNN and Face Net are two
advanced deep learning systems that have received
widespread acclaim for their sophistication and modernity.
Our video streaming layer provides a continuous and stable
experience with no abrupt transitions between consecutive
frames. Fig -1: Face detection

© 2023, IRJET | Impact Factor value: 8.226 | ISO 9001:2008 Certified Journal | Page 128
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 10 Issue: 08 | Aug 2023 www.irjet.net p-ISSN: 2395-0072

The goal of the project is to recognise people in pictures or and other sources could all provide this data. To precisely
videos, extract their face traits, and compare them to a identify registered users, the system will use face recognition
database already in existence. The main goal is to substitute algorithms. Here, faces are found and recognized using the
a dependable and practical solution for manual attendance Haar Cascade method.
methods in order to lighten the administrative burden and
increase accuracy. The project's overall goal is to provide a
seamless, user-friendly system that enhances attendance
management and provides a dependable, effective, and
secure way to record and track attendance.

1.3 LITERATURE REVIEW

Despite notable recent developments in the field of face
recognition [DeepFace, DeepId2], scaling up face verification
and recognition effectively poses substantial obstacles to
existing methodologies.

1) The [1] FaceNet system, which we introduce in this

study, directly learns a mapping from face images to
a condensed Euclidean space where distances
directly correlate to a measure of face similarity.[5] Fig -2: motion detection window
Eigenfaces: This method captures major facial
changes for recognition by using primary The technique of identifying and recognizing faces of people
Component Analysis (PCA) to represent faces as a who are not yet known or registered in a database is known
collection of primary components. as "unknown face detection." It entails examining the
patterns and features of the face to identify whether a face
2) Local Binary Patterns (LBP): A texture-based
resembles any recognized individuals or if it is completely
method for encoding local pixel patterns, LBP
unidentified. The use of object detection algorithms will be
captures spatial relationships and adapts to changes
used to identify and remove unidentified faces from
in lighting.
attendance records. Applying various algorithms and
3) Fisherfaces: Also referred to as Linear Discriminant approaches to analyses, alter, and extract valuable
Analysis (LDA), this method derives distinguishing information from visual data is the process of processing
characteristics from the faces of various people. images and videos. The suggested system will analyses facial
features, conduct face recognition, and find unidentified
4) Deep Convolutional Neural Networks (CNN): By faces using image and video processing techniques. The
extracting intricate patterns and features from facial practice of monitoring and documenting people's presence
photos, CNNs, a subset of deep learning techniques, or absence in a certain setting is known as attendance
have achieved astounding success in face management. The suggested system's attendance
recognition. management process includes organizing the attendance
records, storing timestamped photos of unidentified faces,
5) 3D Face Recognition: This method uses depth data and maintaining a database of recognized people.
from 3D face scans or specialized sensors to enable Attendance management is the process of keeping track of
accurate identification even when position, lighting, and recording people's presence or absence in a certain
and expression variations are present. location. A database of recognized people is kept,
timestamped photographs of unidentifiable faces are stored,
2. METHODOLOGY and attendance records are organized as part of the
suggested system's attendance management procedure.
The suggested method seeks to create a reliable face
recognition attendance system that can detect and subtract 2.1 DATASET
unknown faces while reliably identifying recognized faces.
Unknown faces will be found in the collected photos or video We'll build a database of every person, their unique ID, and
frames using object detection methods. Unknown people will the quantity of grayscale photos needed for face recognition.
have their faces and a timestamp recorded by the system, A dataset is created using the photographs of the specific
which will serve as a record of their existence. The system person, and the images are then stored in an XML file. For
would require input from photos and videos in order to each person's ID, we used 1000 samples because this
distinguish faces and compare them to a database of increased the accuracy of the image of the individual. The
recognizable individuals. CCTV cameras, mobile cameras, authorized individuals' created dataset is kept in a YML file.

© 2023, IRJET | Impact Factor value: 8.226 | ISO 9001:2008 Certified Journal | Page 129
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 10 Issue: 08 | Aug 2023 www.irjet.net p-ISSN: 2395-0072

The images are trained using the Local Binary Pattern in turn. In order to lessen computing effort, the cascade is
Histogram (LBPH) approach. made to swiftly dismiss regions that are unlikely to contain
faces. A region is processed further for face recognition or
Chi-Square: This formula is used to characterizes a person's additional analysis if it successfully navigates through every
expression. In our project, it is employed to assess the stage of the cascade and is deemed to have made a positive
character of a certain individual whose database is recorded detection. After face detection, further methods for face
in an XML file. recognition can be applied. In order to identify the person,
these algorithms examine the identified face region, extract
facial characteristics like landmarks or descriptors, and
contrast them with a database of recognized faces.

Euclidean Distance: To calculate the separation between two

straight lines, use this equation. It is employed to calculate
the dimensions of images that are stored in datasets.

Fig -3: Block diagram of face recognition system

Due of its effectiveness and precision, the Haar Cascade

method is frequently used for face detection tasks. It is
appropriate for use in real-world applications due to its
Normalized Euclidean Distance: The length of the line capacity to manage changes in lighting conditions, facial
segment connecting points hist1 and hist2 (hist1, hist2) angles, and partial occlusions. In complex circumstances with
equals the Euclidean distance between them. A Euclidean extreme position fluctuations or low-resolution photos, it
vector is a representation of a point's location in Euclidean might, nevertheless, be limited. However, for face detection
n-space. This formula scaled the image's length and created a and recognition in computer vision, the Haar Cascade method
square box around it. continues to be a key tool.

Absolute Value: After being acquired by the camera, the

image is trained using both positive and negative images
using the Haar Cascade Classifier. It is employed to change
grayscale photographs into colored ones.

2.2 DESIGN
The technology uses a camera or webcam to record facial
photos or video frames. These images are used as raw
material for later processing. Preprocessing techniques are
applied to the collected images or frames to improve quality
and lower noise. This might entail filters, normalization, and
resizing actions.

The Haar Cascade algorithm employs a series of classifiers

that are staged in a cascade. Each stage consists of a number
of weak classifiers that analyses various areas of the image Fig -4: Flow chart of face recognition system

© 2023, IRJET | Impact Factor value: 8.226 | ISO 9001:2008 Certified Journal | Page 130
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 10 Issue: 08 | Aug 2023 www.irjet.net p-ISSN: 2395-0072

2.3 IMPLEMENTATION However, particular information about these elements is not

given. Future Needs and Improvements: a) Improved
The software configuration is the setting where the Accuracy from Different Camera orientations: Future
project was developed. It is important to choose a language research should concentrate on improving facial recognition
that is suited for the project. Additionally, the right operating accuracy, particularly from various camera orientations. This
system for the project needs to be chosen. These choices have would strengthen the system and enable it to recognize faces
a significant impact on how well the created tool works. taken from different perspectives. b) GPS Location
Therefore, these should only be chosen after a thorough Recording: By including GPS location recording as a feature,
examination. the system's actions would be given more context and
information, which could be valuable for tracking and
The experiment was conducted on a computer running monitoring. c) other Features: The report refers to the need
Windows 11 that featured an Intel i7 4 GHz processor, 8 GB of for other features, but it does not list them. To make the
memory, and a webcam (HP TrueVision HD camera with 0.31 system more usable, additional requirements and desirable
MP and 640x480 resolution). functionalities should be determined.

2.4 RESULT
Numerous strengths and shortcomings of the project
system were discovered during the development and testing
phase. In order to increase the usability of the system, other
enhancements and requirements were also listed. Here is an
overview: System Strengths: a) High Accuracy: The project
system's facial recognition algorithm often displays high
accuracy. This shows that the system does a good job of
accurately recognizing faces. b) No Prior Training Is
Necessary: The facial recognition algorithm employed in this
project system does not require pre-training, in contrast to
several others. This suggests that the system doesn't require
substantial training to quickly adjust to new faces. Systems'
shortcomings a) Camera orientation Dependency: Aspects
like camera orientation have an impact on face recognition
accuracy.
Fig -6: face detection of an known person

We want to create a reliable and effective system to identify

faces in pictures and videos with this face detection project.
Modern deep learning techniques, particularly Convolutional
Neural Networks (CNNs), which have demonstrated
outstanding performance in image identification tests, will
be used. A variety of photos with faces in various stances,
lighting settings, and backgrounds will make up the dataset
for training and validation.

Fig -5: Records automatically when an unknown

person is detected

Fig -7: attendance details

© 2023, IRJET | Impact Factor value: 8.226 | ISO 9001:2008 Certified Journal | Page 131
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 10 Issue: 08 | Aug 2023 www.irjet.net p-ISSN: 2395-0072

To improve the data's generalizability, we will preprocess it [2] Security enhancement utilizing motion detection, Signal
by shrinking, normalizing, and augmenting it. The dataset Processing, Communication, Computing and Networking
will be used to fine-tune the CNN architecture we have Technologies (ICSCCN), 2011 International Conference
chosen, such as the well-known Faster R-CNN or Single Shot on, T huckalay, 2011, pp. 552–557. T homas, Ashraf, Lal,
MultiBox Detector (SSD). For implementation, Python and Mathew, Jayashree.
deep learning frameworks like TensorFlow or PyTorch will
be used. Metrics will be used for evaluation. [3] Moghaddam, B., Pentland, A.P.: Face recognition using
view-based and modular eigenspaces. In: Automatic
3. CONCLUSIONS Systems for the Identification and Inspection of
Hu mans, vol. 2277, pp. 12–22 (1994),Sawhney, S.,
The results of the face detection and face recognition Kacker, K., Jain, S., Singh, S. N., Garg, R. (2o19, January).
systems show that they are both quite successful and Real-time smart attendance system using face
precise. Many different difficulties in daily life can be solved recognition techniques. In 2o19 9th international
using techniques comparable to face recognition technology, conference on cloud computing, data science
which has many applications. In the system we've suggested, engineering (Confluence) (pp. 522-525). IEEE
accuracy is roughly 85%.High accuracy in recognizing people
is one of the main benefits of a face recognition attendance [4] Kodali, R.K., Jain, V., Bose, S., Boppana, L.: IoT based
system. The method enables accurate identification by smart security and home automation system. In: 2016
examining distinctive facial patterns and traits, which leaves International Conference on Computing, Communication
no room for fraudulent activities like proxy attendance. A and Automation (ICCCA), pp. 1286–1289. IEEE (2016)
fair and open environment is fostered by this, which
encourages accountability and integrity in attendance [5] M. Turk and A. Pentland, “Eigenfaces for Recognition,” in
tracking. This method involves locating and detecting faces Journal of cognitive neuroscience, vol 3, pp. 71-86, Jan
in live videos or still photos. In order to ensure that, this calls 1991.
for the adoption of technology and speedier algorithms. To
ensure that the system can handle the data quickly, quicker [6] Roqueiro, Petrushin, Counting people using video
algorithms and technology must be implemented. For use in cameras, Department of Computer Science, University of
the future, every single database entry is updated. Any Illinois at Chicago, Chicago, IL 60607, USA, 2006.
advancement can be implemented into this system because
[7] ] Kartik, J.S., Kumar, K.R., Srimadhavan, V.S.: Security
it has a modular architecture. The system may also include
system with face recognition, SMS alert and embedded
changes to the environment. This system efficiently
network video monitoring terminal. Int. J. Secur. Priv.
completes daily chores using contemporary, in-demand
Trust Manag. (IJSPTM) 2, 15–17 (2013)
technology. To sum up, a face recognition attendance system
has the power to revolutionise attendance monitoring by [8] ] Chin, H.: Face recognition based automated student
offering precise, secure, and effective solutions. Adopting it attendance system. Diss. UTAR (2018)
can improve security, expedite administrative procedures,
and foster an environment of openness and accountability. [9] Ramanan, D., Zhu, X.: Face detection, pose estimation,
This technology can considerably improve attendance and landmark localization in the wild. In: Proceedings of
management in numerous sectors with careful the 2012 IEEE Conference on Computer Vision and
implementation and privacy controls in place, leading to Pattern Recognition (CVPR), pp. 2879–2886 (2012)
overall operational efficiency and productivity
[10] Yue, Wei, and Li. A Video Sequence Segmentation
ACKNOWLEDGEMENT Algorithm for Fore ground Back ground. China’s
Jiangnan University.2015.
The task has been successfully completed, and I am very
appreciative to God Almighty for his grace and blessings, [11] Chu Y, Ahmad T, Bebis G, Zhao L (2017) Low-resolution
without which this study would not have been possible. face recognition with single sample per person. Signal
Process 141:144–15.
REFERENCES
[12] Joint face detection and alignment using multi-task
[1] FaceNet: A unified embedding for face identification and cascaded convolutional net works, K. Zhang, Z. Zhang, Z.
clustering, F. Schroff, D. Kalenichenko, and J. Philbin, Li, and Y. Qiao
2015 IEEE Conference on Computer Vision and Pattern
Recognition (CVPR), pp. 815-823.M. Young, The
Technical Writer’s Handbook. Mill Valley, CA: University
Science, 1989.

© 2023, IRJET | Impact Factor value: 8.226 | ISO 9001:2008 Certified Journal | Page 132