Access Control Using Facial Recognition with Neural Networks for

Restricted Zones

Rodrigo Rea

no, Piero Carri

on and Juan-Pablo Mansilla

Universidad Peruana de Ciencias Aplicadas, School of Engineering, Lima, Peru

Keywords:

Facial Recognition, Access Control, Neural Networks, Artiﬁcial Intelligence, Facial Recognition System.

Abstract:

A new technology that has proven to be effective and accurate in identifying people today is facial recognition.

This technology, when used with IP cameras, provides a very effective and practical access control system.

Moreover, this system is able to learn and improve its facial recognition capability over time through the use of

neural networks, leading to higher accuracy and a lower false positive rate in the ﬁeld. Thus, this paper shows

a face recognition system, based on neural networks, for monitoring and controlling access of people in small

and medium-sized enterprises (SMEs); with the use of IP cameras for the versatility of continuous tracking to

people circulating in restricted areas. On the other hand, common security problems that are identiﬁed in these

environments are addressed and solutions are offered through the implementation of the proposed system.

Finally, the results obtained demonstrate that the system offers an efﬁcient and secure solution for monitoring

and controlling access of people in restricted areas of small and medium-sized enterprises (SMEs). Its accurate

identiﬁcation capability, combined with the elimination of barriers and convenience for users, signiﬁcantly

improves security and user experience.

1 INTRODUCTION

Today, facial recognition has diverse applications in

ﬁelds such as security, surveillance, commerce and

healthcare. It is used to identify and authenticate indi-

viduals by capturing and analyzing unique facial char-

acteristics. However, there are also privacy and se-

curity concerns regarding access control in the afore-

mentioned areas. Access control in restricted areas is

a constant concern for enterprises. The need to protect

information, goods and people from unauthorized or

potentially dangerous access has led to the implemen-

tation of security measures, such as the use of access

cards, security keys, surveillance cameras and secu-

rity guards. However, there are problems that can af-

fect the effectiveness of these controls, such as unau-

thorized access, human error, lack of technology, lack

of training and lack of maintenance.

There is research that uses facial recognition for

other applications such as (Lee et al., 2020) in health-

care, who propose a video surveillance system to

monitor and predict the behavior of patients using

machine learning, which is a branch of artiﬁcial in-

telligence. Likewise, there is a study by (Talahua

et al., 2021) that describes a facial recognition sys-

tem that can identify people with and without a mask

in real time; in addition, there are system proposals,

oriented to the diagnosis of diseases, as described by

(Pan et al., 2021) that performs an automatic facial

recognition system based on deep learning to help in

the diagnosis of Turner syndrome in patients. On the

one hand, (Nyein and Oo, 2019) propose a classroom

attendance registration system using face recognition

and SVM machine learning technique to improve the

efﬁciency of the attendance registration process in

university classrooms. On the other hand, (Xu et al.,

2021) propose a check-in system, in hotels, that uses

facial recognition in order to realize a fast, secure and

private way of check-in.

There are facial recognition systems that are fo-

cused on the student sector (attendance system in

classrooms), health (system that identiﬁes when a per-

son has a mask or not), among others; however, access

control is a difﬁculty encountered by companies and

there is no solution. Thus, this project will not only

focus on the business environment, but to all sectors,

since the proposed software could also be able to be

implemented in banks (biometric control) and airports

(people search).

In this study, a facial recognition system using

neural networks is proposed for access control in re-

stricted areas that offers high accuracy in the identiﬁ-

310

Reaño, R., Carrión, P. and Mansilla, J.

Access Control Using Facial Recognition with Neural Networks for Restricted Zones.

DOI: 10.5220/0012185800003584

In Proceedings of the 19th International Conference on Web Information Systems and Technologies (WEBIST 2023), pages 310-318

ISBN: 978-989-758-672-9; ISSN: 2184-3252

cation of people. In addition, this system uses IP cam-

eras where the facial characteristics of a person are

analyzed and compared with a database of faces pre-

viously stored with the use of neural networks, also

alerts are sent to the assigned users about their respec-

tive area.

The article follows a structure of ﬁve main chap-

ters. Chapter I, Introduction, provides the context of

the study and presents the problem to be addressed.

The objectives of the study are established and the

relevance of the topic is highlighted. Chapter II, Re-

lated Articles, includes a review of the literature and

highlights previous work related to the research topic.

Chapter III, Methodology, describes the project pro-

posal and the algorithm used for face recognition.

Chapter IV, Proposed Solution, details the solution

or methodological approach used in the study; it also

shows the physical and logical architectures of the so-

lution. Chapter V, Results, presents the validation of

the project, along with the costs and the optimality of

the application with respect to other solutions. Chap-

ter VI, Conclusions, discusses its implications and

contributions to the ﬁeld of study. In addition, Chap-

ter VII, Recommendations, offers suggestions for fu-

ture research and possible areas of improvement. This

organizational structure provides a logical and coher-

ent presentation of the study, from the introduction to

the results obtained, to provide a clear and complete

understanding of the research conducted.

2 RELATED ARTICLES

There are articles that are related to facial recognition

recognition as presented by (Hussain and Al Balushi,

2020) that has an effective solution for real-time facial

emotion classiﬁcation using a deep learning model,

with a classiﬁcation accuracy of 89.76 percent. Also

found were (Carlos-Roca et al., 2018) and (Rahmat

et al., 2019) that present a recognition oriented sys-

tem using machine learning techniques to detect and

verify the identity of passengers through a camera and

a car security system that uses facial recognition tech-

niques to verify the identity of the driver respectively.

It is important to mention that the project will focus

on restricted areas, therefore we rely on the proposed

by (Sridhar Chakravarthy et al., 2020) that a secu-

rity access control system that uses facial recognition

techniques to verify the identity of people entering a

restricted area.

On the one hand, there is another line of articles

that perform studies of access control using facial

recognition such as the study performed by (Lopez-

Lopez et al., 2021) where he proposes an approach

for facial recognition that uses a limited dataset and

a deep learning model that is trained autonomously

to improve the accuracy of a real-time facial veriﬁ-

cation system, as well as (Junquera-S

anchez et al.,

2021) in access control architecture that provides a

more complete and adaptive solution for access man-

agement in security systems and can help prevent

more sophisticated security threats. There are also

studies where access control systems are applied as

performed by (Mendez et al., 2021), which uses fa-

cial recognition techniques to verify the identity of

people entering a restricted area; (Lee et al., 2020)

and (Lee, 2021), who propose an access control sys-

tem using facial recognition that can be used in stan-

dalone access control systems and propose a system

with facial detection and recognition algorithms for

identity veriﬁcation of people trying to access a re-

stricted area; (Abou Loume et al., 2022), who present

a facial recognition system that uses a deep learn-

ing algorithm and a cloud service to identify a per-

son and grant access to a door. In the other hand, we

ﬁnd articles where access control using facial recog-

nition converges with the use of neural networks.

For example, what is described by (Almabdy and

Elrefaei, 2019) and (Elmahmudi and Ugail, 2019)

where they describe different approaches based on

convolutional neural networks (CNN) for face recog-

nition, including feature extraction and image classi-

ﬁcation techniques, and a new deep face recognition

method that uses imperfect facial images to improve

recognition accuracy (Araujo et al., 2018). Follow-

ing this idea, (Salama AbdELminaam et al., 2020)

and (Dur

an Su

arez, 2017) proposed a face recogni-

tion system that uses computational intelligence algo-

rithms and deep learning techniques to improve face

recognition accuracy. In addition, there are propos-

als using face recognition, using neural networks, for

the creation of real-time face recognition system ca-

pable of recognizing faces even when wearing masks,

proposed by (Kocacinar et al., 2022) and a system us-

ing cameras embedded in medical devices to capture

facial images and verify the user’s identity before al-

lowing access to the device for patient monitoring via

IoT, proposed by (Hussain et al., 2022) and (Hussain

and Al Balushi, 2020).

3 METHODOLOGY

The development of this project was based on the neu-

ral network methodology developed by AWS Rekog-

nition. Neural networks are a fundamental compo-

nent in the ﬁeld of deep learning, which is a branch of

machine learning. These networks are composed of

Access Control Using Facial Recognition with Neural Networks for Restricted Zones

311

layers of interconnected nodes that simulate the func-

tioning of the human brain. Each node, or artiﬁcial

neuron, processes information and performs mathe-

matical operations on the input data. In addition,

AWS Rekognition uses convolutional neural networks

(CNNs) for computer vision tasks, such as object and

face detection, as well as face recognition and feature

extraction. CNNs are particularly well suited for im-

age analysis because of their ability to learn patterns

and visual features at different levels of abstraction.

It is important to mention that convolutional neu-

ral networks are a deep learning architecture specially

designed for computer vision tasks. These networks

are composed of multiple layers, including convolu-

tional layers, clustering layers and fully connected

layers. Convolutional layers are responsible for ex-

tracting important visual features, such as edges, tex-

tures and patterns, through the convolution of ﬁlters

in the image. These features are then used to classify

and recognize visual elements in the images. (IBM,

2022)

Therefore, the project aims to develop and im-

plement a system that allows the control and moni-

toring of people in restricted areas within small and

medium enterprises. The proposed technological so-

lution is a face recognition system that will be used

by the security area within small and medium com-

panies, given its ﬂexibility for indoor environments.

This real-time face recognition system will be based

on the identiﬁcation of people inside a restricted area

and will validate through facial recognition if these

people are within a whitelist (i.e., previously regis-

tered in the database), otherwise alerts will be sent to

those responsible for internal security (via Email and

WhatsApp). The development will use components

of Cloud Computing, AWS Rekognition as mentioned

above, for the processing of images in real time where

access control will be performed in these restricted ar-

eas.

4 SOLUTION

4.1 Facial Recognition Application

Using AWS Rekognition

Facial Registration and Comparison with AWS

Recognition: This section presents an easy-to-

use face recognition application that uses the AWS

Rekognition service for face registration and com-

parison. The application consists of two main steps:

face registration and face matching for user valida-

tion. The general procedure and the graph of the steps

that make up the solution (Figure 1) is as follows:

Figure 1: Project architecture.

Face Registration: In the face registration process,

images of the user’s face are registered to the user.

These images are sent to AWS Rekognition, which

performs face detection and analysis. AWS Rekogni-

tion extracts facial features from the images and cre-

ates a unique face template for each registered user.

The face template contains facial landmarks and fea-

ture descriptors that represent the user’s face.

Face Matching and Alert Generation: During the

user validation process, the application captures the

camera image and sends it to AWS Rekognition for

face matching. AWS Rekognition compares the cap-

tured face with the registered face templates stored in

its database related to the area assigned to that cam-

era. A similarity score is calculated between the two

faces based on their facial features. If the similarity

score exceeds a predeﬁned threshold of 70%, the ap-

plication considers that the user’s face is a match. In

this case, no alert is generated. However, if the simi-

larity score is below the threshold, indicating a possi-

ble discrepancy, the application generates an alert via

AWS Simple Notiﬁcation Service (SNS). The alert is

sent via mail and WhatsApp message to a designated

administrator or security personnel, notifying them

that there is an unauthorized person.

On the other hand, it is important to show the type

of architectures that were used within the develop-

ment of the project, in addition to show the interface

of the web application of the facial recognition sys-

tem. In this way, the following points are shown to

reinforce the one described:

4.2 Physical Architecture

The physical architecture represents all the physical

components of the system that meet the needs of the

logical architecture and thus allow the correct op-

eration and deployment of the project. The phys-

ical components, such as servers, resources, hard-

ware, that play a role in the technological solution are

shown, as well as the relationship between these com-

ponents, as shown in Figure 2.

WEBIST 2023 - 19th International Conference on Web Information Systems and Technologies

312

Figure 2: Project physical architecture.

4.2.1 Component Description

Internal Devices

• IP Cameras: These are video cameras that are de-

signed to send audio and video signals over the

Internet from a speciﬁc router.

Users

• Security Controller y Personal Security: SME

users who will be able to use the system and/or

receive alerts from unknown visitors.

Internet Connection

• Router: Router that enables local connections

within the company and provides access to the In-

ternet.

• Internet: A network of computers that are con-

nected by the Internet, worldwide, in the form of

a spider’s web.

Interfaces

• Smartphone: Smart cell phone where notiﬁcations

of alerts will be sent to users via WhatsApp or

email.

• Browser client: Browser that will access the client

of the technological solution.

Data

• Azure SQL Server: Microsoft Azure service that

provides a database with the SQL Server engine.

This Microsoft Azure service was chosen because

it is much smaller than other services, according

to Microsoft Azure (2022).

Frontend

• AWS Amplify: Amazon Web Services (AWS)

resource that will contain the ReactJS frontend

(client) of the technological solution.

Backend

• AWS Lambda: Amazon Web Services (AWS) re-

source that will contain the backend in .NET Core

(Web API) of the technological solution.

Algorithm Server

• AWS Rekognition: Amazon Web Services recog-

nition service. This system has the ability to rec-

ognize faces in images and videos. Through this,

it is possible to obtain details about the location

where a face was recognized in an image or video,

as well as information about the position of the

subject’s eyes and any detected emotions (e.g., a

happy or sad expression).

Notiﬁcation Server

• AWS SNS: Amazon Simple Notiﬁcation Ser-

vice is a messaging service that is managed

application-to-application (A2A) and application-

to-person (A2P), as reported by (Services, 2022a).

Security Rules

• The security rules that will protect the integrity,

conﬁdentiality and availability of user and system

data.

4.3 Logical Architecture

This section presents the description of the logical ar-

chitecture, which represents all the layers of the sys-

tem that satisfy the needs of the logical architecture

and, in this way, allow the correct operation and de-

ployment of the technological solution of the project,

as shown in Figure 3 below.

Figure 3: Project Logic Architecture.

4.3.1 Component description

External Items

• IP Cameras: These are video cameras designed

to send video and audio signals over the Internet

from a router.

– Drivers: Drivers for compatibility with other

external and internal devices.

– USB Ports: USB ports for direct connection to

other devices.

Access Control Using Facial Recognition with Neural Networks for Restricted Zones

313

Web Application

• Azure SQL Server: Microsoft Azure service that

provides a database with the SQL Server engine.

This Microsoft Azure service was chosen over

Amazon Web Services because of the price which

is much lower, according to (Azure, 2022).

– SQL Server Database: Database where all sys-

tem information and facial recognition analysis

will be stored.

– SQL Server Instance: Database instance.

– SQL Server: SQL language that will allow data

queries.

• AWS Amplify: Amazon Web Services (AWS)

resource that will contain the ReactJS frontend

(client) of the technological solution.

– Monitoring: Monitoring module within the

technology solution that will allow reviewing

the core business indicators.

– Security Rules: Security rules that protect

the conﬁdentiality, integrity and availability of

data.

– Functions: Functions implemented within the

development that allow to achieve the stated ob-

jective.

• AWS SNS: Amazon Simple Notiﬁcation Ser-

vice is a fully managed messaging service for

application-to-application (A2A) and application-

to-person (A2P) communication, according to

(Services, 2022a).

– Email / SMS Reminder: Internal service of the

resource for sending notiﬁcations.

Users

• Personal Security y Security Controller: SME

users who will be able to use the system and/or

receive alerts from unknown visitors.

Algorithm Server

• AWS Rekognition: Amazon Web Services service

for facial recognition. According to (Services,

2022b), Amazon Rekognition can detect faces in

images and videos, as well as obtain information

about where faces are detected in an image or

video.

– Face Recognition API: API that will allow fa-

cial recognition by comparing two photos.

– Deep neural network models: Model used to

perform face recognition.

User Interfaces

• Browser Client: Browser that will access the

client of the technological solution.

– Web App: Customer of the technological solu-

tion.

• Smartphone: Intelligent cell phone where alerts

notiﬁcations will be sent to users by SMS or e-

mail.

– Message App: Cell phone messaging applica-

tion.

– Email App: Cell phone email application.

5 RESULTS

In this section, you can see the results of the project

from the validation that was performed by testing

the web application of the facial recognition system

where neural networks were used for access control

in restricted areas; in addition, the costs presented by

this project are shown and ﬁnally how feasible its de-

velopment is and how optimal this development be-

comes.

To validate the operation of the web application, a

case study was conducted in May 2023 in a restricted

area of the district of San Miguel, in the city of Lima,

Peru. The important points are shown below:

5.1 Case Study

The web application was validated in an area that is

considered a ”restricted zone” in the district of San

Miguel - Lima, Peru, where we focused on facial

recognition of people whose age ranges from 22 to

45. In these restricted areas, there is a total of 0 to 1

person who are located and perform their activities; in

addition, the access control will be performed in case

a person who does not belong to that area is identi-

ﬁed and also, that the camera does not recognize the

face of the identiﬁed person. Regarding clothing, we

test that people have 2 or more clothes to validate the

correct functioning of our application.

5.2 Standards Adaptation

In the validation, it will be important to develop our

tests in places whose lighting is optimal, although it is

not determinant since our web application recognizes

without depending on this factor. Some of these rules

are presented below:

R01. If the person IS registered (previously) in

the database, the IP camera placed in the restricted

area will show a green bordered box signifying that

the person is authorized to be in the area.

R02. If the person is NOT (previously) registered

in the database, the IP camera placed in the restricted

WEBIST 2023 - 19th International Conference on Web Information Systems and Technologies

314

area will show a box with red borders meaning that

this person is not authorized to be in the place; you

will also receive by e-mail and WhatsApp the mes-

sage ”1 unknown face(s) was identiﬁed in the ABC

zone of the T ﬂoor”.

5.3 Experimentation

The validation was carried out with a sample of 17

people, where the sample was grouped into 3 differ-

ent groups. The ﬁrst group consisted of 6 people; the

second, of 7 people; and ﬁnally the third group, of

4 people. Facial recognition in these zones was car-

ried out in a period of 1 minute, during which time

we waited until the alert message was received by the

owner (authorized person) of the restricted zone. It is

important to note that the participants in this valida-

tion were asked for their consent for the use and ma-

nipulation of data in the development of this project.

Regarding the metrics, the following question-

naire with 8 questions, which are shown below, was

made to the group of people who use the facial recog-

nition web application in order to obtain results that

show us if our application meets the solution to the

identiﬁed need. On the other hand, this question-

naire was carried out in a form made in Google

Forms, which can be found in the following link

https://forms.gle/CmwnK21xEUqC4Tzn6:

Questions

Q1. Is the facial recognition web application easy

to use?

Q2. Do you think the facial recognition web ap-

plication is scalable?

Q3. Do you consider the facial recognition web

application to have a clear and easy to understand user

interface?

Q4. Is the facial recognition web application intu-

itive?

Q5. Does the facial recognition web application

meet industry security standards?

Q6. Is the facial recognition web application ac-

curate in identifying people?

Q7. How often do you consider using the facial

recognition web application to monitor and control

access to restricted areas?

Q8. Would you recommend the facial recognition

web application to other users to control and monitor

access control in restricted areas?

The answers to these randomized questions fol-

lowing the Likert scale (0: not at all, 1: very little, 2:

a little, 3: moderately, 4: a lot). (Joshi et al., 2015)

5.3.1 Indicators

False Positive and False Negative Rate: False posi-

tive rate refers to the number of people who were reg-

istered within the database; however, they are not rec-

ognized by the cameras and show the alert, besides

enclosing the face with the red box (1). On the other

hand, false negative rate refers to the users previously

registered but the IP camera encloses the face with

green color (2).

It should be noted that these values are obtained

per image. As a minimum value, it was proposed

that the rate should not be higher than 5%, thus, we

would be getting closer to the efﬁciency presented by

the web application.

FalsePositiveRate =

(1)

Where:

a: Number of unregistered but identiﬁed persons

(red box).

b: Total number of people identiﬁed in an image.

FalseNegativeRate =

(2)

Where:

c: Number of people registered but not identiﬁed

(green box).

d: Total number of people identiﬁed in an image.

Average Alert Response Time: The alert re-

sponse time corresponds to the indicator that gives us

the subtraction of the time in which the alert was sent

to the user (either by mail or WhatsApp) and the time

in which the alert was created (3). It is important to

mention that the internet speed and other factors that

could affect the delay of this alert are not considered.

It should be noted that these values are obtained by

image.

The value determined as a limit is set at a time no

longer than 10 seconds, so that, if this value is ex-

ceeded, it would not be optimal or beneﬁcial for the

user.

AlertResponseTime = TimeA − TimeB (3)

Where:

TimeA: Alert sending time in seconds.

TimeB: Alert creation time in seconds.

Average Face Recognition Response Time: The fa-

cial recognition response time was found by subtract-

ing the time in which the image processed by the cam-

era was received by the cloud from the time in which

the capture was created in the web application (4). It

Access Control Using Facial Recognition with Neural Networks for Restricted Zones

315

should be noted that these values are obtained per im-

age. As with the previous indicator, it was established

that the minimum allowable value (time) is no more

than 8 seconds, so that our web application is useful

and very functional.

Face Recognition Response Time = TiempoC - TiempoD

(4)

Where:

TimeC: Image reception time in seconds.

TimeD: Alert creation time in seconds.

5.3.2 Results

Corresponding to the results obtained in the project,

it is detailed that before performing facial recognition

in the web application, the user was previously reg-

istered in the database (see Figure 4) so that he/she

could be identiﬁed.

Figure 4: Prior registration of the user to the web applica-

tion.

Previous to this registration, a ﬁrst validation test

was performed, as shown in Figure 5, to demonstrate

that the user was neither identiﬁed nor previously reg-

istered in the face database and the web application is-

sued an alert showing the face of the strangers inside

the restricted area.

After that, a ﬁrst validation was performed, in the

same restricted area, with the same group of people

(with the same characteristics as detailed in the sec-

tion on experimentation) where it recognizes the reg-

istered user (green box) and does not recognize the

other participants (red box) since they were not pre-

viously registered and the camera performs a correct

facial recognition, as shown in Figure 6.

On the other hand, the results obtained from the

usability surveys, with the questions shown in the pre-

Figure 5: Test of alert sending by unregistered user.

Figure 6: Image of ﬁrst project validation.

vious table, show a great acceptance of the web appli-

cation by the 15 users who participated in the valida-

tion.

The results were based on showing the following

information (see Figure 7) because the question was

considered a vital indicator within the project; since,

whether or not they would recommend the web appli-

cation alerts, in summary form, how viable, effective

and secure the application is for the users who used

the web application.

The graph above shows that for question 8, the to-

tal number of people who consider that they would

recommend the application ”Very much” is 66.7%

(equivalent to 10 people), while 20% (3 people) con-

sider that they would recommend the web application

”Moderately”; ﬁnally, the remaining 13.3% consider

that they would recommend the application ”Not very

much”.

WEBIST 2023 - 19th International Conference on Web Information Systems and Technologies

316

Figure 7: Question 8 pie chart.

6 CONCLUSIONS

The facial recognition system based on neural net-

works in IP cameras for access control of people

in SMEs offers an effective and secure solution to

strengthen security in facilities. Through the accurate

and rapid identiﬁcation of individuals, this system im-

proves access control efﬁciency and reduces the risks

associated with unauthorized entry. However, it is im-

portant to note that the successful implementation of

this system requires a proper needs assessment and

careful selection of IP cameras and face database.

In addition, supplementing the system with addi-

tional security measures and performing regular mon-

itoring and maintenance to ensure its proper func-

tioning over time is a way to enhance the ﬁnal pro-

posed solution, as the development of this project

does. Therefore, 66.7% of the people who partici-

pated in the validation are willing to recommend the

facial recognition system for access control.

7 RECOMMENDATIONS

Before implementing the system, it is important to

evaluate the needs and problems that may be encoun-

tered in the sector to which the solution is to be pro-

vided. On the one hand, it is essential to select high

quality cameras, with adequate resolution and image

capture capacity to ensure accurate identiﬁcation; it is

also important to ensure optimal integration with the

facial recognition software, as well as its use.

On the other hand, it is essential to perform regular

monitoring and maintenance of this system to ensure

its proper functioning, which involves verifying the

calibration of the cameras, performing software up-

dates, monitoring the database of faces and reviewing

the access logs to detect any anomaly or attempted

security breach in the sector in which this project was

oriented. Finally, it is necessary that future research

be based on this project since it can be oriented not

only to the SME sector, but to any sector where the

most successful solution is a new facial recognition

system, also the sample size can be larger, in order to

increase the metrics that beneﬁt the development of

the project.

REFERENCES

Abou Loume, G., Abana, A. B., Tonye, E., and Kabiena,

Y. (2022). Facial recognition in the opening of a door

using deep learning and a cloud service. International

Journal of Intelligent Systems and Applications in En-

gineering, 10(3):40–45.

Almabdy, S. and Elrefaei, L. (2019). Deep convolutional

neural network-based approaches for face recognition.

Applied Sciences, 9(20):4397.

Araujo, A., P

erez, J., and Rodriguez, W. (2018). Apli-

caci

on de una red neuronal convolucional para el re-

conocimiento de personas a trav

es de la voz. In

Proc. Sexta Conferencia Nacional de Computaci

on,

Inform

atica y Sistemas, pages 77–81.

Azure, M. (2022). Azure SQL Database: servicio de base

de datos en la nube administrado — Microsoft Azure

— azure.microsoft.com. https://azure.microsoft.com/

es-es/services/sql-database/campaign/#pricing.

Carlos-Roca, L. R., Torres, I. H., and Tena, C. F. (2018).

Facial recognition application for border control. In

2018 International Joint Conference on Neural Net-

works (IJCNN), pages 1–7. IEEE.

Dur

an Su

arez, J. (2017). Redes neuronales convolucionales

en r: Reconocimiento de caracteres escritos a mano.

Elmahmudi, A. and Ugail, H. (2019). Deep face recogni-

tion using imperfect facial data. Future Generation

Computer Systems, 99:213–225.

Hussain, S. A. and Al Balushi, A. S. A. (2020). A real time

face emotion classiﬁcation and recognition using deep

learning model. In Journal of physics: Conference

series, volume 1432, page 012087. IOP Publishing.

Hussain, T., Hussain, D., Hussain, I., AlSalman, H., Hus-

sain, S., Ullah, S. S., and Al-Hadhrami, S. (2022). In-

ternet of things with deep learning-based face recog-

nition approach for authentication in control medical

systems. Computational and Mathematical Methods

in Medicine, 2022.

IBM (2022). ¿Qu

e son las redes neuronales convolu-

cionales? — IBM — ibm.com. https://www.ibm.com/

es-es/topics/convolutional-neural-networks.

Joshi, A., Kale, S., Chandel, S., and Pal, D. K. (2015). Lik-

ert scale: Explored and explained. British journal of

applied science & technology, 7(4):396.

Junquera-S

anchez, J., Cilleruelo, C., De-Marcos, L., and

Martinez-Herr

aiz, J.-J. (2021). Access control be-

yond authentication. Security and Communication

Networks, 2021:1–11.

Kocacinar, B., Tas, B., Akbulut, F. P., Catal, C., and Mishra,

D. (2022). A real-time cnn-based lightweight mo-

bile masked face recognition system. Ieee Access,

10:63496–63507.

Access Control Using Facial Recognition with Neural Networks for Restricted Zones

317

Lee, H., Park, S.-H., Yoo, J.-H., Jung, S.-H., and Huh, J.-

H. (2020). Face recognition at a distance for a stand-

alone access control system. Sensors, 20(3):785.

Lee, H.-W. (2021). Design of multi-functional access con-

trol system. IEEE Access, 9:85255–85264.

Lopez-Lopez, E., Regueiro, C. V., Pardo, X. M., Franco, A.,

and Lumini, A. (2021). Towards a self-sufﬁcient face

veriﬁcation system. Expert Systems with Applications,

174:114734.

Mendez, D. F., Molina, M. V., Forero, M. G., and Lugo, C.

(2021). Facial recognition system for security access

control. In Applications of Digital Image Processing

XLIV, volume 11842, pages 386–400. SPIE.

Nyein, T. and Oo, A. N. (2019). University classroom

attendance system using facenet and support vector

machine. In 2019 International conference on ad-

vanced information technologies (ICAIT), pages 171–

176. IEEE.

Pan, Z., Shen, Z., Zhu, H., Bao, Y., Liang, S., Wang, S.,

Li, X., Niu, L., Dong, X., Shang, X., et al. (2021).

Clinical application of an automatic facial recognition

system based on deep learning for diagnosis of turner

syndrome. Endocrine, 72:865–873.

Rahmat, R., Loi, M., Faza, S., Arisandi, D., and Budiarto,

R. (2019). Facial recognition for car security system

using ﬁsherface method. In Journal of Physics: Con-

ference Series, volume 1235, page 012119. IOP Pub-

lishing.

Salama AbdELminaam, D., Almansori, A. M., Taha, M.,

and Badr, E. (2020). A deep facial recognition system

using computational intelligent algorithms. Plos one,

15(12):e0242269.

Services, A. W. (2022a). AWS — Servicio de no-

tiﬁcaciones Push (SNS) — aws.amazon.com.

https://aws.amazon.com/es/sns/?whats-new-cards.

sort-by=item.additionalFields.postDateTime&

whats-new-cards.sort-order=desc.

Services, A. W. (2022b). Detecting and analyzing faces

- Amazon Rekognition — docs.aws.amazon.com.

https://docs.aws.amazon.com/rekognition/latest/dg/

faces.html.

Sridhar Chakravarthy, G., Anupam, K., Harish Varma, P.,

Teja, G. H., and Rodda, S. (2020). Face recognition

with voice assistance for the visually challenged. In

Intelligent Computing and Communication: Proceed-

ings of 3rd ICICC 2019, Bangalore 3, pages 701–709.

Springer.

Talahua, J. S., Buele, J., Calvopi

na, P., and Varela-Ald

as, J.

(2021). Facial recognition system for people with and

without face mask in times of the covid-19 pandemic.

Sustainability, 13(12):6900.

Xu, F. Z., Zhang, Y., Zhang, T., and Wang, J. (2021). Fa-

cial recognition check-in services at hotels. Journal

of Hospitality Marketing & Management, 30(3):373–

393.

WEBIST 2023 - 19th International Conference on Web Information Systems and Technologies

318