Microsoft Word - 83----58 (636 - 642 Page) (B)---(사우디 Copyright Accepted) 0904 Mushir An Arabic Edutainment Application

IJCSNS International Journal of Computer Science and Network Security, VOL.22 No.9, September 2022

636

Manuscript received September 5, 2022

Manuscript revised September 20, 2022

https://doi.org/

10.22937/IJCSNS.2022.22.9.83

Mushir: An Arabic Edutainment Application For Deaf and Hard of

Hearing Children Using Real-time Image Processing

Haya Alnasif

, Lujain Alyahya

, Haifa Alromaih

, Ghada Alhelal

, Nada Barnawi

, Arwa Altamim

Sarah Albassam

King Saud University, Information Systems Department, Riyadh, Saudi Arabia

Summary

Education is an essential stage of children lives. Children with special nee

ds (deaf and hard hearing) may encounter challenges and need more assist

ant in their education journey. Recent technology could be utilized to impr

ove the deaf and hard hearing children learning experience. For these reaso

ns, this paper proposed “Mushir” an Arabic application that targets deaf an

d hard of hearing children from four to seven years old. "Mushir" aims to i

ncrease child knowledge and enthusiasm for learning letters, numbers, sha

pes, and colors in Arabic sign language. In addition, the child can examine

the knowledge he achieves by playing two different games. The first gam

e is a multiple-choices questions. The second game utilized image processi

ng technology by applying real-time processing for the camera senses. In t

his game the child will be asked to open the camera and indicate a movem

ent of his/her hand in sign language.

Keywords:

Real-time image processing, deaf and hard of hearing, children, Arabic

Application

1. Introduction

As we all know, "Education in childhood is like

engraving on stone," so one of the most important things that

must be considered for a child's life is education. With

education, our nation will rise and evolve. Preschool

education is a remarkable stage full of energy and

excitement that should be considered by deaf children's

parents; learning the basics at this stage might positively

affect the children in advanced stages [1].

The dilemma is that education opportunities differ from

child to child, especially for people with special needs (deaf

and hard of hearing) because they have much fewer learning

opportunities than healthy kids do, so they need more care

in their education and future. Technology changed

everything; since the internet started spreading, learning has

become different, so we have to take advantage of it for

people with special needs. Not being attentive to this part of

society (children with special needs deaf and hard of hearing)

caused a lack of educational applications for children with

special needs in Arabic sign language. Most of the

applications available focused on communication purposes

or apps that educate people in sign language [2]. Therefore,

to support their educational journey and encourage them to

learn an educational and entertaining application is proposed.

Mushir aims to support deaf and hard of hearing children

in learning basics such as letters, numbers, shapes, and

colors in Arabic sign language (ARSL) using image-

processing techniques to make the application more

interactive. Arabic sign language (ARSL) is a full natural

language that used by the deaf in Arab countries to

communicate in their community. Unfamiliarity with this

language increases the isolation of deaf people from society.

This language has a different structure, word order, and

lexicon than Arabic. The translation between ARSL and

Arabic is a complete machine translation challenge because

the two languages have different structures and grammars

[3].

2. Literature Review

2.1 Related Studies

Technology plays an important role in our society,

especially in the field of education. It is quite regrettable that

children, particularly the deaf, still face many challenges in

acquiring an education. Their learning methods are different

as compared to hearing people. They use Sign Language (SL)

rather than natural language to communicate and learn [4].

In this literature review, we are going to discuss the

difficulties that deaf and hard of hearing children face and

how technology can make a huge difference in educating

them. In the work presented in [5], Al-Megren and Almutairi

elicited the user requirements for the development of an

augmented reality (AR) application that supports the literacy

development of Arab children who are hearing impaired.

The findings indicated that the parents and teachers

preferred Arabic SL (ARSL), pictures, and videos, whereas

the children struggled with ARSL and preferred

fingerspelling. These preferences highlighted the

importance of integrating various resources to strengthen the

written Arabic and ARSL literacy of Arab children. It also

showed the importance of establishing requirements elicited

directly from intended users to support their learning process.

The study results were used in the preliminary development

of Word & Sign, an AR mobile application intended to aid

IJCSNS International Journal of Computer Science and Network Security, VOL.22 No.9, September 2022

637

Arab children who are hearing impaired in their linguistic

development.

Komal Parvez et al [4] developed a mobile application

for learning basic mathematical concepts using Pakistan

Sign Language (PSL), and the objective was to determine

the effectiveness of the application. The study bridged the

gap between the technology-based method and conventional

teaching methods. The sample was 192 deaf participants

aged 5–10 years. They were divided into two groups; one

group learned through conventional methods (flashcards and

board) and the other group through the developed mobile

application. The difference in the performance of both

groups was evaluated by conducting quizzes. The findings

revealed that the Experimental Group (EG) participants,

who were instructed by the mobile application showed

higher proficiency in the quizzes as compared to the Control

Group (CG). CG participants took 20 min longer than EG

participants to complete the quizzes. The results showed that

the quiz scores were directly affected by the mode of

teaching used for participants in both groups.

Bouzid et al [6] aimed to examine the deaf learners' interest

in using an educational game for learning the sign language

notation system Signwriting. The results found indicated

that, overall, the application is useful, enjoyable, and easy to

use. The game can stimulate the students' interest in learning

such as notations. Alnafjan et al [7] describe the design and

the development of an iOS application. The application

provides the chance for children with difficult hearing, ages

7-12, to learn about basic Islamic tenets like ablution (Al-

Wudu), daily prayer (Salah), and supplication (Al-Adhkar)

using the Arabic Sign Language (ARSL).

2.2 Related Applications

 SignLang : is an iOS application (iPhone and iPad)

designed for all ages who want to learn about the

American Manual Alphabet (AMA). It uses real-time

image processing hand shape and motion trackers to

recognize sign language. The application has a simple

interface with two sections, and there is an option to

choose the left or right hand. the training section shows all

letters in the sign language at the bottom of the screen. The

user can select any letter to see it as an alphabet and try to

shape it with his/her hand using the camera and it shows

the user the percent of matching to the letter. In the

challenge section, the user will start the challenge when

entering the section with a timer of one minute and they

have to answer questions as possible as they can [8].

 ASL Kids : Is an application that works on IOS and

Android systems provide a set of American Sign

Language (ASL) dictionary to help deaf and hard of

hearing children to learn and examine their knowledge. It

includes fingerspelling letters where the child can see the

letters and its corresponding hand sign. In addition, when

a child tests his/her knowledge, a gif picture is provided

for sign language movements. In addition, a multiple

choice quizzes are provided. The application teaches

children signs about various subjects like family, colors,

animals, etc. An audio button option is available if needed

to hear how a word is pronounced to stimulate speech and

hearing. [9].

 Hands on ASL : Is an application available for both iOS

and Android. It aims to teach children the communication-

basics using the American Sign Language alphabet

interactively uses 3D hand models that can be rotated from

all angles. The application targets children from the age of

4 and up. It provides quizzes of varying difficulties and

allows learners to choose the level to test their knowledge.

[10].

 Baby sign dictionary : Is an application works on IOS and

Android systems. Edutainment is provided in a set of

different options, chosen based on the child's interests

such as work words, animals, daily routines, feelings and

emotions, food, morals, behavior, and nature - all arranged

alphabetically for quick and easy searches. Moreover, the

child can see an image, replay a video tutorial for each

sign, and learn sign language movement with a voice

where it describes instructions on sign language

movement and how to do it in detail. Furthermore, the

application also comes with a fun interactive video quiz

and voice button if needed. [11].

 Educational letters for deaf : Is an educational application

for Android targets deaf aged 6-8 and enables them to

learn the Signal alphabet in several entertainments’ ways

using multimedia, such as attractive colors, graphics, and

shapes. It also includes simple games that enable children

to understand the letters with an example. The app

supports Arabic and English [12].

2.3 Discussion

Early childhood learning of sign language for deaf and

hard of hearing children results in increasing their knowledge

and improving their confidence. Mushir application will help

Deaf and hard of hearing children to be able to communicate

freely with everyone, learn easily in school and to be

involved more in the society.

The applications reviewed above raised our knowledge

and awareness of the most important features and

characteristics that should considered in edutainment

applications, such as attractive design and variety of displays.

The reviewed applications have many features, but they also

come with limitations. Some of the most common limitations

are: many of these applications do not support Arabic

language, lack of a reward system and they do not support

image processing. In contrast, Mushir application provides

all these features.

3. Problem Definition

Many people in the Arab world suffer from various

disabilities and difficulties. In Saudi Arabia, the General

Authority for Statistics stated in 2017 that people with

IJCSNS International Journal of Computer Science and Network Security, VOL.22 No.9, September 2022

638

hearing difficulties reached 1.4% of its population [13], and

this category constitutes a large proportion of sign language

users. In particular, the number of Saudis aged five and over

who use sign language is (27748) person. There is no doubt

that children make up a large part of this group.[14]

Children with disabilities in our Arab world, specially the

deaf and hard of hearing, may face some challenges in their

educational journey. Like other children aged 4-7 years, they

need to be prepared to enter school. They are in the stage of

having some basic knowledge of letters, numbers, shapes,

colors, etc.

The educational process is witnessing a lot of progress and

development. Technology and entertainment in our time are

an essential part of education [15]. Children with speech and

hearing disabilities lack Arabic entertainment educational

programs that meet and emulate their needs [2].

Our application Mushir aims to provide suitable means

of education that contribute effectively to deaf and hard of

hearing children’s’ education. Also, to develop children's

capabilities and experiences and prepare them to join their

school community.

The application consists of four categories the child can

learn from which are: letters, numbers, shapes, and colors.

Each of these categories has two sections, learn lesson and

play game sections. Thus, the child learns and then tests

what has been learned through an entertaining game. The

child interacts with the game using image processing or by

choosing the correct answer. In each of the four categories,

there are stages to motivate the child to overpass these stages

and complete the game. In addition, the application allows

the parent to add their emails to receive periodical feedback

about their child progress.

Our project's primary goal is to develop an Arabic

educational entertainment application for deaf and hard of

hearing preschoolers to teach them learning basics such as:

letters, numbers, shapes, and colors. In a comfortable,

interactive, and age-appropriate way by taking into

consideration, their needs. [16].

Main Objectives:

 Help deaf and hard of hearing children to learn the basics

at an early age.

 Enhance children's Sign language in a playful way,

applying learning with fun strategies to raise their

motivations.

 Help parents in supporting the learning of their deaf or

hard of hearing children and follow their progress.

 Integrate the entertainment factor with education to

improve the child's skills and ability to obtain more

knowledge.

 Build a usable, user-friendly application that expanding

children's perception, improving their language,

enhancing their imagination.

4. Application Development

This section shows how we developed the application,

starting with information gathering strategies, design of the

application, user interfaces, implementation and finally

testing the application

4.1 Information Gathering

To get a better understanding of our targeted audience,

we conducted two interviews one with a special-needs

education supervisor, Raed Algofily, and the other with a

certified trainer in Arabic sing language Khaloofa Alshehri,

who is also a deaf person. The goal was to understand deaf

children’ needs and utilize modern technology to satisfy

these needs. Thus, improve the outcome of our project.

The interview revealed that often deaf children lack the prior

knowledge about letters and numbers, if they were not enroll

in kindergarten unless there are interest and initiative from

their parents, and this is rare among deaf children. In

addition, we wanted to know how technology can contribute

to the education of deaf and hard of hearing children, and we

got an interesting answer. Technology reduces burden and

time due to children's need for casual learning, the

technology makes it easier for parents who are not skilled in

sign language to teach their children sign language as well

as learning it themselves.

Besides the interviews, we have conducted a

questionnaire using Google Forms, targeting the parents of

deaf and hard-of-hearing children who speak Arabic in

Saudi Arabia and 195 responded to our survey. The responds

we received inspired us to be creative in designing Mushir

to meet the children needs.

In particular, the survey results highlighted the importance

of educational application in raising children cognitive level.

It also indicated the parents’ preference to have the

educational materials presented in an entertaining way. The

results also confirm the shortage of an interactive Arabic

application targeting deaf children.

Finally, the questionnaire participants mentioned some

features and suggestions for the application to be more

attractive and motivational for the children. Some of these

features are: the ease of use, interaction by adding voice

feature to the application and providing the content in Arabic

language. We considered those suggestions in our

construction and development of the application.

4.2 System Design

Mushir is designed as an educational and entertaining

application that helps deaf and hard of hearing children aged

IJCSNS International Journal of Computer Science and Network Security, VOL.22 No.9, September 2022

639

from four to seven years learning basics sign language. The

application applies image-processing techniques to make the

application more interactive and it is in Arabic language. As

the application targets children, the design and the

application interfaces are very important. The theme we have

applied is “sky & clouds”, which will add flamboyance to the

application and a bird character designed to guide the child

throughout the application. The application consist of four

categories the child can learn from which are: letters,

numbers, shapes, and colors. Each of these categories has two

sections, learn lesson and play game sections. Thus, the child

learns and then tests what has been learned through an

entertaining game. The child interacts with the game using

image processing or by choosing the correct answer. In each

of the four categories, there are stages to motivate the child

to overpass these stages and complete the game. In addition,

the application allows the parent to add their emails to receive

a periodical feedback about their child progress.

Mushir is an application based on real-time image

processing technology, so the proposed architecture is client

and server architecture. In three-tier client-server

architectures, there are three main layers located on three

different machines: user interface, application server, and

database server. Each of these layers has specific roles and

responsibilities. The first layer is the user interface, which is

the layer where the users would have direct interaction with

the application. The second layer is the application server

which is the processing level acting as a bridge to connect the

application logic with the database servers' data. The third

layer is the database server the data level, has the primary

responsibility of storing all the data in a database server.

Since the layers separate the logical elements from the

physical servers, it would be much easier to be maintained.

Mushir application is based on client and server architecture

to deal with storage. It requires information from an external

source that works as a server, which is the image recognition

API, which helps obtain sufficient information for the

functionality that the application performs.

4.3 User Interfaces

This part presents a sample of screen shots in Mushir

application to illustrate the user interface design (Fig. 1 to 10).

Fig.8. Play Game with

"Image Processing”.

Fig.3. Home Page, a Nice

Home Page that Contains

the Cate

Fig.4. Select section

Page, Appears after the

user Choose Cate

Fig.2. Registration Page,

the user has to Enter his

Information

Fig.5. Learn lesson Page,

A List according to the

selected category Appears

for the user to Choose

from.

Fig.6. Lesson Page,

Where the Child Learns

and Understand the

Necessar

Skills.

Fig.7. Stages Page, a

Bunch of Different Stages

in the Play Game Section.

Fig.9. Play Game by

Choosing the Answer.

Fig.10. Child

progress, Shows the

Child's

IJCSNS International Journal of Computer Science and Network Security, VOL.22 No.9, September 2022

640

4.4 System implementation

Implementation of Mushir developed in an Android

Studio environment with Java programming language. Java

was chosen since the team was familiar with the language and

because it includes multiple resources to help in

implementing functions. The application interfaces

implemented using Figma, which is a cloud-based UI and UX

design application that can be used to create websites and

apps.

To run the application and ensure the functions work

properly, Mushir team used an Android virtual device

provided by Android Studio and Samsung mobile. Moreover,

Mushir is connected to a real-time Firebase database that

store users’ information and category content. The Firebase

Realtime Database is a cloud-hosted NoSQL database that

lets you store and sync between users in real-time. Therefore,

the exchange of data to and from the database is easy and fast.

In the beginning, the application starts to run the login

page, so the user can log in or register if he/she does not have

an account. Then the user will reach the home page, which

contains four categories, and they are divided into (letters -

numbers - shapes - colors), through which the child can

choose. The child can choose the play section or the learning

section. The child can learn from the content of the category

he/she has chosen. In addition, the child can play in two

different ways the question can be answered by either

opening the camera then indicating the pose of his/her hand

in sign language using real-time image processing

technology, or through the multiple-choice answers.

4.5 Image processing implementation

The implementation of Mushir application began by

building an Arabic sign language detector using Python

libraries and TensorFlow object detection API that is a

foundation library that can be used to create Deep Learning

models [16]. First, we start by cloning a real-time object

detection repository from GitHub to leverage transfer

learning and the training code inside of the repository. Then,

we collect images using OpenCV (Open-Source Computer

Vision Library) which is an open-source computer vision and

machine learning software library [17]. Editing Python code

was by using Jupyter Notebook, which allows us to create

and share documents that contain live code [18].

Then we label the images by using the LabelImg package,

which is a free open-source tool for graphically labeling

images written in Python to draw detection boxes against the

different sign language poses. After labeling the images, we

split them into training and testing partitions using the

Holdout method, which is randomly partitioning the data into

two independent sets for training, and testing [19].

Then we generate two TensorFlow records for training

and testing, then we train the deep learning model in

command promote. After the model trained, we test the

trained model with the webcam called Jupyter Notebook.

After the model worked properly, we change the file format

from TensorFlow to TensorFlow lite which is a set of tools

to help developers run TensorFlow models on mobile to

allow us to link the model into Android studio and integrate

it with the rest of the code [20].

4.6 System evaluation and testing

We perform system testing to uncover the errors in our

system and ensure that our application performs accordingly

to our specification. First, we begin with the unit testing

where each component in the system is tested individually.

After that, we conduct the user acceptance testing in the real-

world scenarios with our targeted users.

User Acceptance Testing is a phase of testing where the

system is being tested by targeted end-users to determine

whether the software can be accepted or not. We conducted

our test on 4 Arabic deaf and hard of hearing preschool

children and one parent for a child, their age ranges from 4 to

7 years old. The table below shows the User Acceptance

Testing results and the given feedback.

Table 1: User Acceptance Testing

Aljwharah bint Abdullaziz

Task

Number

Errors

Time

needed

User

feedback

Completion

status

Signup 0 00:58:90 Pass

Reset

password

0 00:19:44 - Pass

View

account

0 00:14:58 - Pass

Edit

account

0 00:25:54 - Pass

Select

category

0 00:13:08 - Pass

Learn

lesson

0 00:55:19 - Pass

Play

simulation

game

0 00:20:03 - Pass

Open

camera

0 00:09:08 - Pass

Process

gesture

0 00:23:37 - Pass

Play

match

game

0 00:24:48 - Pass

Choose

answer

0 00:12:44

ﺢﺻ ﷲ"

"!ﻲﻠﻋ

“wow it

correct”

Pass

Play voice 0 00:09:05 - Pass

IJCSNS International Journal of Computer Science and Network Security, VOL.22 No.9, September 2022

641

View hint 0 00:01:50

ﻭﻮﻠﺣ"

ﺕﻮﺼﻟﺍ

"ﺢﺿﺍﺍﻭ

“nice

and the

voice is

clear”

Pass

View

progress

0 00:22:58 - Pass

progress

by email

0 00:40:03

ﻩﻭﻭﺃ "

ﺔﻧﺍﺎﻨﻓ ﻲﻠﻜﺷ

ﻱﺬﻬﺑ

"!ﺔﺒﻌﻠﻟﺍ

“Oh it

seems

that I am

a genius

of this

game”

Pass

Logout

0 00:02:04 - Pass

The test results indicates that the application functions

worked as expected. The users found Mushir application

very educative, enjoyable, and useful. Also, they liked play

simulation game the most.

5. Conclusion

In this paper, we proposed Mushir an Arabic application

that targets deaf and hard of hearing children from the age of

four to seven years old. The application consists of four

categories that enables the child to learn from: letters,

numbers, shapes, and colors. Each of these categories has two

sections, learn lesson and play game sections. Thus, the child

learns and then tests what has been learned through an

entertaining game. The child interacts with the game using

image processing or by choosing the correct answer. In each

of the four categories, there are stages to motivate the child

to overpass these stages and complete the game. In addition,

the application allows the parent to add their emails to receive

periodical feedback about their child progress. We seek to be

the best choice for all parents of the deaf and hard of hearing

children, and that we enrich the Arabic content for pre-school

applications for deaf and hard of hearing children.

6. Future Work

One of the challenges Mushir faced was finding people to

make interview with due to COVID-19 restrictions. So it took

time to find the right people. Mushir application have only

four categories to learn and play games and support only a

single game player. Also, its currently available only to

Android users, and due to these limitations, Mushir team

looking forward expanding the application to cover more

categories like family, food, nature, animals etc. Also, the

application will support two or more game players and will

be compatible with the iOS environment.

References

[1] Clement, S., Digital Learning:Education and skills in the digital age,

Rand.org, 2017. [Online]. Available:

https://www.rand.org/content/dam/rand/pubs/conf_proceedings/CF3

00/CF369/RAND_CF369z1.arabic.pdf. [Accessed: 23- Sep- 2020].

[2] AlImam, M., AlAbadi, Z., The Effectiveeness of a Program Based on

Electronic Games in developing Creative Thinking Skills for Deaf

Children, Educational and Psychological Studies, vol. 66, pp. 262-207

(2010) [Online]. Available:

http://search.mandumah.com/Record/111117. [Accessed: 23- Sep-

2020].

[3] Mohandes, M., Automatic Translation of Arabic Text to Arabic Sign

Language, AIML Journal, vol. 6 (4), pp. 15-19 (2006) [Online].

Available:

https://www.researchgate.net/publication/325782734_Automatic_tra

nslation_of_Arabic_text-to-Arabic_sign_language. [Accessed: 21-

Jan2021].

[4] Parvez, K., Khan, M., Iqbal, J., Tahir, M., Alghamdi, A., Alqarni, M.,

Alzaidi, A., Javaid, N., Measuring Effectiveness of Mobile

Application in Learning Basic Mathematical Concepts Using Sign

Languag, Sustainability, vol 11(11), (2019).

[5] Al-Megren, S., Almutairi, A., User Requirement Analysis of a Mobile

Augmented Reality Application to Support Literacy Development

Among Children With Hearing Impairments, Journal of Information

and Communication Technology, (2019).

[6] Bouzid, Y., Khenissi, M., Essalmi, F., Jemni, M., Using Educational

Games for Sign Language Learning - A SignWriting Learning Game:

Case Study, Educational Technology and Society, vol. 19 (1), pp. 129-

141 (2016). Available:

https://www.jstor.org/stable/jeductechsoci.19.1.129?seq=1.

[Accessed 30 September 2020].

[7] Alnafjan, A., Alaskar, H., Aljumaah, A., Alshraihi, R., Designing

“Najeeb”: Technology-Enhanced Learning For Children With

Impaired Hearing Using Arabic Sign-Language ArSL Applications,

International Conference on Computer and Applications (ICCA), pp.

238-273(2017). [Online]. Available:

https://ieeexplore.ieee.org/abstract/document/8079772/metrics#metri

cs. [Accessed: 02- Oct- 2020].

[8] SignLang, App Store. [Online]. Available:

https://apps.apple.com/us/app/signlang/id1414850534. [Accessed:

02- Oct- 2020].

[9] ASL kids, App Store [Online]. Available:

https://apps.apple.com/us/app/asl-kids-fun-sign-

language/id962456700. [Accessed: 02- Oct- 2020].

[10] PlayNice, Playnice.games, 2020. [Online]. Available:

https://playnice.games/handsonasl.html. [Accessed: 02- Oct- 2020].

[11] Baby Sign and Learn - Baby Sign Language Apps using ASL

(American Sign Language), Babysignandlearn.com, 2020. [Online].

Available: http://www.babysignandlearn.com/asl-dictionary-app.

[Accessed: 02- Oct- 2020].

[12] " ﻖﻴﺒﻄﺗ ﻑﻭﺮﺤﻟﺍ ﺔﻴﻤﻴﻠﻌﺘﻟﺍ ﻢﺼﻠﻟ ", Play.google.com. [Online]. Available:

https://play.google.com/store/apps/details?id=air.Character.educatio

n.for.the5.deaf. [Accessed: 01- Oct- 2020].

[13] Elm, Twitter.com, 2020. [Online]. Available:

https://twitter.com/elm/status/1099619628353875968?lang=ar.

[Accessed: 23- Sep- 2020].

[14] General Authority for Statistics, Disability Survey, 2020. [Online].

Available: https://www.stats.gov.sa/en/904. [Accessed: 23- Sep-

2020].

[15] Rushdi, S., The Reality of the Use of the Deaf and Hard of Hearing

Students of Instructional Technology in light of some of the Variables

in Higher Education, Journal of Special Education and Rehabilitation,

vol. 5(20), pp. 66-113 (2017) [Online]. Available:

http://search.mandumah.com/Record/835954. [Accessed: 23- Sep-

2020].

[16] Brownlee, J., Introduction to the Python Deep Learning Library

TensorFlow, Machine Learning Mastery, 2021. [Online]. Available:

IJCSNS International Journal of Computer Science and Network Security, VOL.22 No.9, September 2022

642

https://machinelearningmastery.com/introduction-python-deep-

learning-library-tensorflow/. [Accessed: 20- Apr- 2021].

[17] About - OpenCV, OpenCV, 2021. [Online]. Available:

https://opencv.org/about/. [Accessed: 20- Apr- 2021].

[18] Project Jupyter, Jupyter.org, 2021. [Online]. Available:

https://jupyter.org/. [Accessed: 21- Apr- 2021].

[19] Nelson, J., Getting Started with LabelImg for Labeling Object

Detection Data, Roboflow Blog, 09-Dec-2020. [Online]. Available:

https://blog.roboflow.com/labelimg/. [Accessed: 21-Apr-2021]

[20] TensorFlow Lite guide, TensorFlow, 2021. [Online]. Available:

https://www.tensorflow.org/lite/guide. [Accessed: 21- Apr- 2021].