2022 5th International Conference on Contemporary Computing and Informatics (IC3I)

Advanced Face Detection using Machine Learning

And AI-based Algorithm
2022 5th International Conference on Contemporary Computing and Informatics (IC3I) | 979-8-3503-9826-7/22/$31.00 ©2022 IEEE | DOI: 10.1109/IC3I56241.2022.10072527

Sakshi Sandeep Phatak Harshvardhan Sangram Patil Muhammad Waqas Arshad

Computer Science and Engineering Computer Science and Engineering Department of Creative Technologies
Shivaji University Shivaji University Air University
Kolhapur, Maharashtra, India Kolhapur Islamabad, Pakistan.
sakshis.phatak@gmail.com patilharshvardhan0508@gmail.com muhammad.waqas.arshad.1@gmail.com

Dr. Bhagatsing Jitkar Dr. Sangram Patil Dr. Jaydeep Patil

Department of Computer Science and Department of Computer Science and Department of Computer Science and
Engineering Engineering Engineering
D Y Patil Agricultural and Technical D Y Patil Agricultural and Technical D Y Patil Agricultural and Technical
University University University
Kolhapur, Maharashtra, India Kolhapur, Maharashtra, India Kolhapur, Maharashtra, India
bjitkar@rediffmail.com sangrampatil@dyp-atu.org jaydeeppatil@dyp-atu.org

Abstract- In this study, the analysis of the research that has

been provided in the “Advanced Face Detection using Machine
Learning And AI-based algorithms”. It is a facial
acknowledgment framework that utilizes AI and Machine
learning. It specifically uses the Support Vector Machine (SVM)
system. Modern-day technology is making people amazed with
the innovations that it is providing to make an individual's life
simpler and easier. In the case of Face Recognition, it has proven
to be the least intrusive and also the fastest form of the biometric
verification system. Among all the tasks that can be done with
the help of machine learning algorithms, one of the most crucial
computer vision tasks is the use of face detection and
recognition. The face detection and the recognition systems are
both related to each other yet they are also very different from
one another as well. By using machine learning the face
detection aspect is used and it also is a widespread aspect of the
face recognition system. Machine learning applies various
algorithms for the detection of faces and the recognition process
as well. Face detection and recognition systems are used in
medical diagnosis and also in the deep analysis of human faces Figure 1: System Overview of Face Detection
in the video for various intelligence purposes. Face recognition (Source: Created by Author)
is a sub-category system of the biometric software system, it is
used to map an individual's facial features and also to store that Machine Learning (ML) is a sub-category of Artificial
data as a face print for future purposes. Intelligence (AI) and the field of AI is being covered by a lot
of territories. In simple words, it is a simulation of human
Keywords: Advanced Face Detection, Machine Learning, AI- intelligence that is being used by machines or computers [7].
based Algorithm, and Facial recognition technology Machine learning involves various programming from
various algorithms that can be learned by themselves and can
I. INTRODUCTION also make their predictions, it allows machines to learn from
their past experiences just like humans do. The output that is
Facial recognition technology is fairly ubiquitous in the being achieved by this is then used as input for the next
modern day. The use of the face detection and recognition operation. The Machine Learning algorithms learn from the
system is very omnipresent and many individuals are previously available data to solve the problems that are
unaware of it although it is being used by various individuals currently being faced by conventional programs. Deep
with the help of their smartphones. Advanced face detection learning is also a sub-category of Machine Learning that is
and recognition software are being used by various being derived from the multiple layers of Machine Learning
surveillance operators that can pick criminal faces out of algorithms together at the same time.
crowds [19]. The lesser known factors of this system are the For the past several years a great deal of research has
developments regarding the computer vision and machine been done on identifying and recognizing people. The easiest
learning technology that is happening behind face way to recognize people is the use of Advanced Face
recognition. Detection, because of this reason it has become a hot topic for

1111
979-8-3503-9826-7/22/$31.00 ©2022 IEEE

Authorized licensed use limited to: HKBK College of Engineering. Downloaded on October 07,2024 at 09:43:12 UTC from IEEE Xplore. Restrictions apply.
2022 5th International Conference on Contemporary Computing and Informatics (IC3I)

researchers to engage in the biometric industry [5]. In this

system, various tracking and surveillance approaches have ● To analyze the challenges regarding the Variation of
been implemented that are known as the Landmark. Pose
● To analyze the challenges regarding the Variation in
II. AIM AND OBJECTIVES OF THE STUDY Illumination
● To analyze the challenges regarding the Variation of
Aim: Expression
● To analyse the Issues regarding Ageing
This study aims to analyze the problems that are being ● To analyze the Occlusions issues
faced by the Advanced Face Detection and Recognition ● To analyze the issues regarding the Similarity of
system with the help of Machine Learning and AI-based Faces
algorithms. This study will also provide solutions to the ● To analyze the issues regarding the Resolution of the
problem with the help of a brief description. The biometric Images
software named face detection and facial recognition uses ● To analyze the issues regarding the algorithms of
deep learning algorithms which are a part of machine learning Face Detection
algorithms to compare a live captured image with another
stored face print so that it can verify an individual's identity. III. METHODOLOGY
The backbone of this technology is the processing of the
image and also machine learning. The increased use of facial In this part, secondary data has been collected from
recognition is due to the facts of various human activities that various books, journals, and websites. The data has also been
are being found everywhere [2]. The facial recognition collected from various research that was previously
systems applications are used for security purposes like in the conducted. This data has been provided here with the help of
airport, face tracking, criminal detection, etc. The face various graphs, tables, and charts.
biometric system can be on-intrusive in comparison to other
biometric traits like the palm print, iris, fingerprint, etc.

Figure 2: Algorithms for face recognition Figure 3: Real-Time Face Recognition

(Source: [6]) (Source: [9])

As the use of face detection and recognition can be done IV. RESULTS AND ANALYSIS OF THE STUDY
without an individual's knowledge, the data is used for
security-based applications like criminal detection, airport In the study of this topic, the results that are needed for
security, face tracking, criminal detection, and also in various the challenges regarding the analysis of advanced face
forensic surveillance systems. The face recognition system detection using machine learning and AI-based algorithms
works by capturing an image of a face from video or any have been provided here. The challenges that have been
surveillance camera, and then it is compared with the found by the analysis are-
previously available data that is present inside a database [4].
It involves the training of known images, classifying them A. Challenges regarding the Variation of Pose
according to their class and lastly, they are being stored in the
database. For example, if an image is provided to the system, The issues regarding the variation of Pose create lots of
it will classify and compare that image first then it will store issues in the detection of face. The variation of pose happens
that image inside a database. because of the various changes in the angle of observation of
the observer and can also append due to the rotation of the
Objectives: head position [6]. This variation creates lots of problems in
the identification of the input image.
There is various range of applications of face detection
and recognition system and because of these reasons the
challenges regarding this system is increasing and various
issues of this system are rising [8]. The efficiency of this
system is being challenged at every moment. The various
problems of this study are -

1112

Authorized licensed use limited to: HKBK College of Engineering. Downloaded on October 07,2024 at 09:43:12 UTC from IEEE Xplore. Restrictions apply.
2022 5th International Conference on Contemporary Computing and Informatics (IC3I)

a recognition system needs to detect the different facial

expressions and evaluate the emotional state of that particular
individual [25]. The expressions of human beings consist of
various kinds of macro-expressions such as disgust, anger,
fear, happiness, sadness, surprise, and various other
involuntary and rapid facial patterns.

Figure 4: Variations of Pose

(Source: ijar)

Various systems can tolerate the small variations of the

small rotations in the angles. In the case of large rotational
changes in angles, it becomes very difficult for the system to
detect as the database mostly consists of face images of the
frontal view of the face. Since the beginning of its use, the
face recognition system is very sensitive to the variation of Figure 6: Variation of Expressions
poses [20]. These problems regarding the variations of the (Source: ijrar)
different poses can be solved by the efficient use of various
techniques that are aimed to rotate the face or align the With the help of dense optical flow, the facial changes
image's axis points. can be computer into information. In addition to this, other
things that cause variations in facial expressions like the use
B. Challenges regarding the Variation in Illumination of cosmetics and hairstyles can also be included in this.

The issues regarding the variations of the illuminations D. Issues regarding Ageing
make the efficiency of the face recognition system very
doubtful and questionable[10]. It is very difficult for this
system to perform its task in moderate lighting in various
backgrounds. The variations of illumination can vary as
various objects have different levels of reflection capabilities.
The total magnitude of light intensity that is being reflected
Figure 7: Ageing
from an object can be different in different backgrounds and (Source: ijrar)
because of this reason the face recognition system faces
problems. The issues regarding the Ageing, it is another reason for
the appearance change of the face. This is another major
reason why the efficiency of the face detection and
recognition system is being questioned [11]. As human
beings are biological creations, the ever-changing process of
ageing is continuous. It causes the face detection and
recognition system to provide inaccurate data. If the time
between an image capture is lengthy, then there will be
significant changes in a particular individual. As per various
Figure 5: Variation in Illumination studies that have been done by researchers, it is said that
(Source: iijar) every 10 years there will be significant changes in an
individual's facial appearance. The above mention figure 5
To solve this problem, various kinds of newly modified shows how individuals face changes over time because of
algorithms are being implemented in the system such as the aging [16]. Not only the shape and the lines of a face are
equalization technique. To tolerate the issues regarding the getting modified over time, but the changes in hairstyle are
illumination, several algorithms can be used simultaneously also being noticed as well.
at the time of face recognition, but this system is not fully
reliable so it is advised not to completely trust these E. Occlusions issues
algorithms [13].

C. Challenges regarding the Variation of Expression

The issues regarding the variety of expressions make the
efficiency of the face detection and recognition system very
doubtful. Depending on an individual's state of emotions, the Figure 8: Occlusion
variation of different expressions is being caused. Therefore, (Source: ijrar)

1113

Authorized licensed use limited to: HKBK College of Engineering. Downloaded on October 07,2024 at 09:43:12 UTC from IEEE Xplore. Restrictions apply.
2022 5th International Conference on Contemporary Computing and Informatics (IC3I)

In the issues regarding Occlusion, the different variations environment, the performance quality of the acquiring
of an individual's facial appearance can also happen because system, and various other reasons that can be mentioned as
of the various objects, such as occlusion that are party factors that may be responsible for the variation of the
covering the face. It makes the face detection and recognition resolutions of the image [17].
systems work more troublesome [14].

Figure 11: Variation of Resolutions

(Source: ijrar)

The efficiency of the face detection and recognition

system can be improved with the help of good resolution. So
the resolution of an image is directly linked to the efficiency
Figure 9: Deep Face Recognition of the face detection and recognition system. The solution to
(Source: [11]) this problem is the use of an old technology named the
infrared sensor[26]. The sensor will work even in pitch black
Although the system will be able to find the individual's darkness in the recognition of an individual face. The
faces, it will be difficult for that system to recognize them due implication of the Infrared Sensor in the facial detection and
to the hidden facial parts. It will make the process of recognition system is a must need in the biometric system
recognition very difficult. The challenges regarding these [15].
problems can be seen in real-life scenarios as well, such as a
person talking on the phone, wearing spectacles, hats, scarves H. Challenges regarding the algorithms of Face Detection
or sometimes having their face covered with hands as well
[22]. The modern-day Face Detection and Recognition system
use the state-of-the-art facial recognition method. The
F. Challenges regarding the Similarity of Faces methods rely on a complex and unsuitable architecture named
the Convolutional Neural Network (CNN)[27]. This method
is not suitable for real-time performance on embedded
devices [23]. Other than these various other algorithms are
being used for the face detection and recognition system such
as the Principal Component Analysis (PCA), Independent
Component Analysis (ICA), Linear Discriminant Analysis
(LDA), Elastic Bunch Graph Matching (EBGM), and lastly
Figure 10: Similarity in Faces the Fisherfaces [18].
(Source: ijrar)

In the issues regarding the similarity of faces, it is usually

not a common challenge. Even for humans, it is very difficult
for us to recognize the difference in various similar-looking
faces [21]. Hence, we can imagine how difficult it is for a
computer to identify similar-faced individuals. In the
identification of identical twins, the face detection and
recognition system finds it very problematic as the face
features and shapes do not match [12]. This increases the
False Recognition Rate (FRR) of a particular individual[28]. Figure 12: Algorithms used for face recognition
(Source: [10])
G. Challenges regarding the Resolution of the Images
Ideal face detection and recognition system need to be
In the issue regarding the image resolutions, the quality tolerant of variations in illumination, expression, pose, and
and the resolutions of the image matter a lot. The resolution occlusion. It must be scalable to a huge number of users with
of the image is another important issue where face detection the need to capture minimal images during its registrational
and recognition is being varied with the quality and the time while simultaneously solving the problems regarding the
resolution of the images that are provided as input. Various complex architecture[29].
factors can affect an image's resolutions, such as the
V. CONCLUSION

1114

Authorized licensed use limited to: HKBK College of Engineering. Downloaded on October 07,2024 at 09:43:12 UTC from IEEE Xplore. Restrictions apply.
2022 5th International Conference on Contemporary Computing and Informatics (IC3I)

being used without the knowledge of the users of that device,

As the face of a human being is an essential part of itself, such as the companies like iProov and Mastercard so that they
its unique features make it very crucial in the process of can authenticate the payments and other high-end
identification. To make the face detection and recognition authentications that are required for the tasks regarding
system more advanced, flawless, accurate, and reliable payments.
various kinds of algorithms and technologies are being
implicated in this system[30]. The application of this REFERENCES
biometric system is also expanding in security, healthcare,
forensics, defense, and also in various kinds of [1] Deeba, Farah, et al. "LBPH-based enhanced real-time face
transportations system where accuracy is required. In this recognition." International Journal of Advanced Computer Science
and Applications 10.5 (2019).
analysis, various challenges are being found, such as the
[2] Bah, Serign Modou, and Fang Ming. "An improved face recognition
challenges regarding pose, occlusion, expressions, aging, algorithm and its application in the attendance management system."
resolution of the images, the similarity of faces, and lastly the Array 5 (2020): 100014.
problems regarding the model of algorithms that are being [3] Bah, Serign Modou, and Fang Ming. "An improved face recognition
algorithm and its application in the attendance management system."
used in the face detection and recognition system. The
Array 5 (2020): 100014.
solution to all the challenges has been analyzed and a brief [4] Chakraborty, Chinmay, and Arij Naser Abougreen. "Intelligent internet
discussion of them has also been provided here with the help of things and advanced machine learning techniques for COVID-19."
of various graphs, charts, and images. EAI Endorsed Transactions on Pervasive Health and Technology 7.26
(2021): e1.
[5] Nagrath, Preeti, et al. "SSDMNV2: A real-time DNN-based face mask
VI. ACKNOWLEDGEMENTS detection system using single shot multibox detector and
MobileNetV2." Sustainable cities and society 66 (2021): 102692.
The face detection and recognition system which are a [6] Wang, Zhongyuan, et al. "Masked face recognition dataset and
application." arXiv preprint arXiv:2003.09093 (2020).
crucial part of biometric technologies are one of the safest
[7] Khayyat, Mashael, et al. "Advanced deep learning-based
ways to authenticate your security. The face detection and computational offloading for multilevel vehicular edge-cloud
recognition system are slowly expanding its borders towards computing networks." IEEE Access 8 (2020): 137052-137062.
the retail, travel, public safety, and healthcare industries. In [8] Cui, Feiyun, et al. "Advancing biosensors with machine learning." ACS
sensors 5.11 (2020): 3346-3364.
the healthcare sector, one can expect that various countries all
[9] Wang, Haoyu, and Lulu Guo. "Research on face recognition based on
around the world are adopting face detection and recognition deep learning." 2021 3rd International Conference on Artificial
systems more. It will improve the overall safety standards of Intelligence and Advanced Manufacture (AIAM). IEEE, 2021.
the industry as well as will also reduce identification errors. [10] Loey, Mohamed, et al. "A hybrid deep transfer learning model with
machine learning methods for face mask detection in the era of the
With this, it will also improve the overall customer
COVID-19 pandemic." Measurement 167 (2021): 108288.
environment and experience, such as hospitals and doctor's [11] Patel, Keyur, et al. "Facial sentiment analysis using AI techniques:
surgeries. state-of-the-art, taxonomies, and challenges." IEEE Access 8 (2020):
90495-90519.
[12] Pei, Zhao, et al. "Face recognition via deep learning using data
Face detection and recognition system will be a feature
augmentation based on orthogonal experiments." Electronics 8.10
for the upcoming smart cities. This will increase the safety of (2019): 1088.
the public that will be living in those smart cities. The same [13] Filali, Hajar, et al. "Multiple face detection based on machine
technology principles will also be used in recognizing things learning." 2018 International Conference on Intelligent Systems and
Computer Vision (ISCV). IEEE, 2018.
that are out of place, this principle is known as “object
[14] Sharma, Neha, Reecha Sharma, and Neeru Jindal. "Machine learning
recognition”. and deep learning applications-a vision." Global Transitions
Proceedings 2.1 (2021): 24-28.
The security industry is also turning its head toward the [15] Marsot, Mathieu, et al. "An adaptive pig face recognition approach
using Convolutional Neural Networks." computers and Electronics in
AI-based face detection and recognition system as well. To
Agriculture 173 (2020): 105386.
improve the security around the home and also to improve [16] Khan, Muhammad Zeeshan, et al. "Deep unified model for face
overall access and also to create a more seamless experience recognition based on convolution neural network and edge computing."
the AI-based face detection and recognition system is being IEEE Access 7 (2019): 72622-72633.
[17] Kaur, Paramjit, et al. "Facial-recognition algorithms: A literature
slowly implemented by the security industry. In modern
review." Medicine, Science and the Law 60.2 (2020): 131-139.
times smart home systems, the implementation of AI-based [18] Abdullah, Sharmeen M. Saleem Abdullah, et al. "Multimodal emotion
face detection and recognition system is seen as a security recognition using deep learning." Journal of Applied Science and
upgrade from the previously used biometric systems. Technology Trends 2.02 (2021): 52-58.
[19] Kumar, Ashu, Amandeep Kaur, and Munish Kumar. "Face detection
techniques: a review." Artificial Intelligence Review 52.2 (2019): 927-
The facial recognition solutions are expected to be 948.
present in around 1.6 billion devices by the end of 2024. All [20] Jose, Edwin, et al. "Face recognition based surveillance system using
of them will be powered by AI-based algorithms that will be facet and MTCN on jetson tx2." 2019 5th International Conference on
Advanced Computing & Communication Systems (ICACCS). IEEE,
integrated into the face detection and recognition software in
2019.
mobile phones. In present times the AI-based and machine
learning-based face detection and recognition systems are

1115

Authorized licensed use limited to: HKBK College of Engineering. Downloaded on October 07,2024 at 09:43:12 UTC from IEEE Xplore. Restrictions apply.
2022 5th International Conference on Contemporary Computing and Informatics (IC3I)

[21] Ge, Shiming, et al. "Occluded face recognition in the wild by identity-
diversity inpainting." IEEE Transactions on Circuits and Systems for
Video Technology 30.10 (2020): 3387-3397.
[22] Zulfiqar, Maheen, et al. "Deep face recognition for biometric
authentication." 2019 international conference on electrical,
communication, and computer engineering (ICECCE). IEEE, 2019.
[23] Pranav, E., et al. "Facial emotion recognition using deep convolutional
neural network." 2020 6th International conference on advanced
computing and communication Systems (ICACCS). IEEE, 2020.
[24] Lu, Xiaofeng. "Deep learning based emotion recognition and
visualization of figural representation." Frontiers in Psychology 12
(2021).
[25] Rigby, Michael J. "Ethical dimensions of using artificial intelligence in
health care." AMA Journal of Ethics 21.2 (2019): 121-124.
[26] Ghosh, S., Rana, A., & Kansal, V. A benchmarking framework using
nonlinear manifold detection techniques for software defect prediction.
International Journal of Computational Science and Engineering, 2020,
21(4), 593-614.
[27] Rana, A., & Sharma, S., Mechanism of sphingosine‐1‐phosphate
induced cardioprotection against I/R injury in diabetic rat heart:
Possible involvement of glycogen synthase kinase 3β and
mitochondrial permeability transition pore. Clinical and Experimental
Pharmacology and Physiology, 2016, 43(2), 166-173.
[28] H. Walia, A. Rana and V. Kansal, "A Naïve Bayes Approach for
working on Gurmukhi Word Sense Disambiguation," 2017 6th
International Conference on Reliability, Infocom Technologies and
Optimization (Trends and Future Directions) (ICRITO), Noida, India,
2017, pp. 432-435, doi: 10.1109/ICRITO.2017.8342465.
[29] M. S. Raghavendra, P. Chawla and A. Rana, "A Survey of
Optimization Algorithms for Fog Computing Service Placement,"
2020 8th International Conference on Reliability, Infocom
Technologies and Optimization (Trends and Future Directions)
(ICRITO), Noida, India, 2020, pp. 259-262, doi:
10.1109/ICRITO48877.2020.9197885.
[30] Singh, G., Bhardwaj, G., Singh, S.V., Garg, V. (2021). Biometric
Identification System: Security and Privacy Concern. In: Awasthi, S.,
Travieso-González, C.M., Sanyal, G., Kumar Singh, D. (eds) Artificial
Intelligence for a Sustainable Industry 4.0. Springer, Cham.
https://doi.org/10.1007/978-3-030-77070-9_15.

1116

Authorized licensed use limited to: HKBK College of Engineering. Downloaded on October 07,2024 at 09:43:12 UTC from IEEE Xplore. Restrictions apply.