Jose Rizal University

Drone Surveillance System for Tracking Fire Incidents

A Research Paper/ Thesis Submitted to the

Faculty of the College of Computer Studies and Engineering

In Partial Fulfillment

Of the Requirements for the Degree

Bachelor of Science in Computer Engineering

By

EZK

Dela Cruz, Jabin Clyde

Delos Santos, Ma. Russel Ann

Gamboa, Adrian Z

Morales, Raemon Julian

Midyear, SY 2021-2022

i
 Table of Contents

Preliminaries Page

Title Page ................................................. i

Table of Contents ..................................... ii

CHAPTER 1- INTRODUCTION

1.1 Background of the Study................................ 1

1.2 Statement of the Problem................................ 4

1.3 Objective of the Study...................................…5

1.4 Scope and Limitations.......................................5

1.5 Significance of the Study..................................5

CHAPTER 2 - REVIEW OF RELATED LITERATURE

2.0 Introduction..................................................... 6

2.1 Related Literatures.......................................... 6

2.1.1 Drone for Firefighting....................... 6

2.1.2 Surveillance System.......................... 8

2.1.3 Image and Video Processing............. 9

2.1.4 Object Detection.................................10

2.2 Related Studies.............................................. 11

2.2.1 Fire Monitoring.................................11

2.2.2 Fire Unmanned Aerial Vehicles........12

ii
CHAPTER 3 - RESEARCH METHODOLOGY

3.0 Introduction..................................................... 15

3.1 Research Design.............................................. 15

3.2 Conceptual Framework.................................... 16

3.3 Data Collection Techniques............................. 17

3.4 Block Diagram................................................... 19

3.4.1 System Diagram for Mobile Application…20

3.5 Hardware Design........................................... 21

3.5.1 Components and Materials............. 21

3.6 Software Design............................................ 25

3.7 System Flowchart .......................................... 26

3.8 Concept of Operation..................................... 27

iii
 Chapter 1

INTRODUCTION

1.1 BACKGROUND OF THE STUDY

Over the previous few years, Mandaluyong City has recorded several fire accidents. It
also appears that this catastrophe has occasionally impacted other properties, safety, and
families. People who struggle to live in areas like squatter camps are the ones that suffer the
most whenever a fire occurs. Because of the restricted space for fire trucks to access and
escape, this location. They have to overcome several challenges, which causes the fire truck to
arrive late. According to the Bureau of Fire Protection (BFP) interview, this is also what was said
in the majority of reports. Consequently, the population suffers an excessive number of
casualties. In light of the current state of the covid and the events going on around us, we are all
experiencing difficulty. Due to this, it was particularly difficult when a fire broke out in the
neighborhood.

The speed with which the fire trucks arrive at the hearth location has a considerable
influence on the speed with which fire fighters manage a hearth case. The management of
fireplace cases, which is strongly impacted based on the speed with which the hearth trucks
arrive at the fireplace location, is one of the essential characteristics that may indicate a
firefighter's performance. Travel distance, motor truck speed, fireside case time, traffic
circumstances, and land use type on the route covered are all elements that might impact fire
engine time. As a result, so as to get the fastest route alternatives, an analysis must be
performed to see the fastest route for the firefighter while taking under consideration some
variables. Fires occur often in the Philippines' informal settlements, resulting in death, serious
affliction, as well as the devastation of property and life. Poverty, marginalization, congestion,
and insufficient service availability expose informal settlement members to a variety of risks,
including fires and illness. The literature has focused on the threat of flames in informal
communities. There is less awareness of how informal settlement communities respond to fire
catastrophes, as has the longer-term community reaction to the following destruction. The
firefighters' slow response is one of the most difficult causes of fires to combat and number of
fires increased as a result of the ineffectiveness of the hearth fighting road's operation, which
was exacerbated by traffic conditions that resulted in the huge population of settlement
structures, the combustibility of construction materials, and thus the lack of publicly sponsored

1
firefighting systems. The architectural landscape is constantly changing as new high-rise
structures and complex interior spaces are constructed, which may result in unforeseen
disasters and accidents. As a result, Effective disaster mitigation and relief are required to
ensure the safety of people and property. Firefighting is extremely vital for public and building
safety, especially as fires often occur in low-rise buildings utilized by fire departments. Today is
two-dimensional (2D). As a result, fire commanders previously only had standard fire and
rescue maps that they might use to deploy firefighters. There is a lack of infrastructure
development in informal settlements, notably a lack of access roads. Local communities are
frequently confronted with fire disasters with little support or aid. Even when assistance is
provided, fire prevention equipment is usually delivered too close to the end, just after fire has
destroyed multiple houses or ravaged whole villages. There are two issues with community fire
search and rescue within urban slums. The first concern is how people handle the situation in
the event of a fire and the second is how people will deal with the fallout. The first part is little
understood, while the second is well-documented, with studies on diverse communal coping
mechanisms in the aftermath of a hearth. One type of optimization that affects the transportation
process is the assessment of the fastest or best time.

In a study made by (Yang et al., 2019), In addition to being a crucial component of

preventing forest fires, accurate real-time monitoring is a crucial tool for effectively containing
the spread of forest fires and minimizing associated financial losses. Forest regions are
frequently widely dispersed geographically and difficult to inspect manually. Therefore, in order
to guarantee the security of forest resources, we require a highly mobile early forest fire warning
system. UAVs are used as a highly mobile inspection tool to satisfy the needs of forest fire
inspection. In addition, it is difficult to detect early flame features and it is simple to trigger a
false alarm using single source fire detection. In light of this, we suggest a technique for
detecting forest fires that makes use of optical pictures for smoke detection and infrared images
for fire detection. In this study, a UAV-based approach for detecting forest fires is proposed.
This method conducts fire discrimination using optical and infrared data as opposed to the
earlier single source fire detection method, which improves the robustness of the discrimination
results. In addition, the smoke detection method suggested in this paper uses the frame
difference method. Only when the drone is hovering does this technique work. The method fails
because the drone's flying causes the background to shift quickly. High false positive and false
negative rates continue to be an issue for existing algorithms. Overall, the majority of the
algorithms are highly scenario specific and are particularly sensitive to outside intervention. The

2
algorithm's performance will be impacted if the scenario gets more complicated. According to
(Sungheetha et al., 2020), The study of smart cities and smart infrastructure is currently a
burgeoning area of study. Open spaces, agricultural land, forests, workplaces, residences, and
other places are all susceptible to fire incidents that result in considerable resource losses. In
this research, we combine wireless sensor network and unmanned aerial vehicle (UAV)
technology for early fire detection. This aids in preventing catastrophic fire incidents. The sensor
architecture is used to keep track of the environmental factors. The sensors process the
received environmental data using IoT-based apps. In order to detect fires, this article combines
cloud computing, IoT sensors, wireless technology, and UAVs. The system incorporates image
processing techniques in order to increase the system's accuracy. Fire detection and
extinguishing are major concerns for all countries due to varied infrastructure upgrades and the
implementation of smart cities. When these catastrophes are averted on a worldwide scale,
efficient technology can perform at their peak. The IoT sensors and WSN platform are used for
early-stage fire detection. The World Social Network is hosted on Sensenuts platform. Using the
installed sensor nodes, real-time data gathering and analysis are carried out in either an indoor
or outdoor context. All the collected data is accommodated in cloud storage for later
examination. The ThingSpeak cloud program is used to analyze parameters such as light
intensity, humidity, temperature, smoke, and other variables. Based on effective fire detection
and sensor inputs of the ambient conditions, real-time data is analyzed. Lastly in research by
(Yuan et al., 2017), Unmanned aerial vehicle (UAV) based computer vision systems are
increasingly being used as a more and more promising solution for monitoring and detecting
forest fires. This study presents an image processing technique for the automatic identification
of forest fires in infrared (IR) photographs, with application to UAV. For the purpose of detecting
fire pixels, the presented algorithm uses brightness and motion cues in addition to image
processing methods based on histogram-based segmentation and optical flow approach. First,
the hot objects are extracted as fire candidate zones using histogram-based segmentation. The
motion vectors of the candidate locations are then calculated using the optical flow approach.
To differentiate flames from other fire analogs, the motion vectors are also further examined.
This research suggests a fire detection technique for the use of IR camera-based UAV-based
forest fire surveillance. This method boosts the precision and dependability of forest fire
detection by utilizing both the brightness and motion characteristics of fire in IR images. Fires
from background as well as non-fire hot moving objects can be well separated utilizing
histogram-based segmentation and optical flow analysis. In IR firing video sequences,
experimental validations are carried out; positive experimental outcomes are attained with

3
noticeably better performance. Future research can focus on combining IR and visible range
imagery to decrease the number of false alarms generated when detecting fires.
In a study made by (Yang et al., 2019), non-fire hot zones are still mistakenly recognized
when using infrared images for UAV-based forest fire surveillance because of heat or light
reflected to them. There's a potential that the smoke is preventing the fire from spreading. In IR
photos, smoke can also seem translucent. According to (Sungheetha et al., 2020), Reduced
hardware should be used for real-time monitoring and fire detection using IoT and cloud-based
drones, and sensor networks should be added to detect fire in various situations. Lastly in
research by (Yuan et al., 2017), The Unmanned Aerial Vehicle Platform frame difference
method for early forest fire detection works best when the drone is hovering. When the scene is
disturbed, their algorithm becomes vulnerable. Even if there are more publicly available data
sets, the unified video smoke detection still lacks information.

This research aims to develop a drone to help firefighters to have a vision of how large
the fire is by using image processing through the video that the drone was taking. Extracting the
color of the fire and measuring its area by square meters. Integrating different individual
learners can considerably increase the robustness of the model and boost detection
performance because each individual learner has their own area of expertise and can extract
unique information from the image. To describe the flame features from fire images, the HSI
color model is chosen. The color separation method is used for color segmentation and a fire
mask is applied on original images. The project could help firefighters on how to handle the fire
and make wise decisions on how to stop it from spreading.

1.2 STATEMENT OF THE PROBLEM

A firefighter or fire truck's delayed arrival might result in significantly more harm to the
neighborhood, including the loss of homes and other properties as well as the potential death of
a person. The following research questions were addressed by the proposed project:

● What issues are there with fire monitoring right now?

● Where exactly is the device stationary?
● Why are drones required for this project?
● When a fire occurs, how can the device be mobilized?
● Who will utilize and benefit most immediately from the proposed device?

4
1.3 OBJECTIVES OF THE STUDY

The general objective of the study is to develop a drone monitoring device that can aid
firefighters and government officials in surveying the fire incident and recommending the most
effective route possible for the rescuers.

This study aimed to:

● Create a surveillance drone to help the firemen and to set the scene for the present
crisis.
● Design a fire detection system using image processing.
● Provides information about current fire events via the dashboard.
● Compare the results of the previous studies.
● Test the drone's turnaround time, mistake rate, and false prediction rate, evaluate the
system's effectiveness and accuracy.

1.4 SCOPE AND LIMITATIONS

This section focused on the scope, boundaries, and constraints of the proposed study.
This study included the following features:

● A drone machine that would provide fire fighters with wide vision via a camera.
● A fire detection device that detects flames using images processing.
● A notice that may be accessed via a mobile application to inform the user.

The study, however, was limited to the following

● The image quality of the camera might have had an impact on the fire detection system's
performance.

1.5 SIGNIFICANCE OF THE STUDY

This study’s findings focused on developing a drone that allows firefighters to help them
monitor the fire from above, considering that fire is destructive, destroying homes, and taking
people and animal’s lives. Through this research, it will allow them to be more aware and wiser
in what action they will be making in responding to the fire with the information that they have.

5
The following people in particular will gain from this research:

Firefighters. It will allow them to have a vision of the current situation. It will also be easy for
them to navigate through the streets and possible road blockage with the help of the device.

Officials. They will have information from the controller of the drone on how large and what
location the fire is. It will be easy for them to spread the information to their fellow officials.

Chapter 2

REVIEW OF RELATED LITERATURE AND RELATED STUDIES

This chapter gives a summary of earlier studies on intranets and knowledge sharing.
It introduces the case study framework, which is the main subject of the
research covered in this thesis.

The topic of using drones for real time video to update and to inform everyone what was
happening from above has been a study for a couple of years now especially in countries who
have advanced technology in them. These systems share a number of procedures, but they
vary in how they are carried out. The flow of the system is as follows: (1) real-time video of the
fire; (2) monitoring how they would respond to the fire; (3) analyze the fire.

2.1 Related Literature

2.1.1 Drone for Firefighting

Drones are aircraft that don't have a human pilot, a flight crew, or passengers.
Unmanned aircraft vehicles, which include a controller on the ground and a communications
network for the UAV, include UAVs as a component.
In (IEEE Communications Society & Institute of Electrical and Electronics Engineers,
n.d.) Created a practical concept for deploying many autonomous UAVs to monitor catastrophes
in remote places. The UAVs with longer flight times, greater communication ranges, and more

6
computing power serve as the leaders to form a group of coalitions that can monitor the fire
zone with the expected quality and for the required duration. This monitoring mission is carried
out using a leader-follower coalition formation approach. It is presumably that a number of
unmanned aerial vehicles (UAVs) with various flight and sensor capabilities are stationed at
lookout tower stations or charging in standby mode. The system may therefore be expanded to
deal with circumstances with vast coverage regions, such forest fires.

In the paper by (Nazar Zadeh et al., 2021) At the project's conclusion, an octocopter
unmanned aerial vehicle has been successfully created and put into use. This UAV is
specifically made to shoot and drop class A to class C fire distinguisher balls using PVC
materials. To help them target and shoot the fire extinguisher balls, the firefighting team
employs an HD camera with night vision, real-time video streaming, and a 5.8 GHz radio
frequency (RF) transmission. In this work, the firing and dropping mechanisms are operated by
a remote controller that employs 2.4 GHz RF signals to initiate and regulate the UAV's actions.
With the help of an internal GPS and gyroscope, the drone's flight has been stabilized. The
hardware, shooting, and dropping processes of the UAV were tested throughout this trial, with a
total passing rate of 97.22 percent. One of the challenges faced throughout the test was the
UAV's sensitivity to weather conditions, which might reduce shooting precision at the target site.
As a result, knowledge is required to take control of the UAV's controller.

As described by (Ausonio et al., 2021) Propose a new approach based on the

employment of a swarm of cooperative unmanned aerial vehicles (UAVs) that can transport vast
amounts of fractionated extinguishing liquid and release it on fire fronts, replicating the effects of
rain. This technology is intended to reduce the risk to human life by not only detecting fires but
also putting them out through collaboration with the previously mentioned resources, such as
airborne troops and ground people. This study aims to assess as strongly as possible the
effects of using drones to combat forest fires, a complicated phenomenon that depends on
several variables.

Similar in (Wang et al., 2022) The usage of firefighting drones in wildfires has
increasingly increased as research and technology have advanced. In view of the widespread
problem of firefighting and rescue for high-rise structures, the current study suggests the fire
drone system to put out high-rise building fires from the outside. They use the LY100 system as
an example to describe the application situations and firefighting tactics. Inside this research, a

7
new extinguishing system called the LY100 system is created in light of the challenges
associated with battling fires in high-rise buildings. Its quick deployment and operational
flexibility are its standout benefits. The main components of the LY100 system are the twin-rotor
drone, high-pressure liquid fire extinguishing equipment, pressure fire extinguishing equipment,
vehicle, and extinguishing agent. Additionally, it may work in tandem with elevating fire vehicles
to efficiently contain and put out flames in high-rise buildings. The spray distance of high-
pressure liquid fire extinguishing equipment and the spraying of dry powder by pressure fire
extinguishing equipment have both been used to illustrate the efficacy of the proposed LY100
fire extinguishing system.
In the study of (Aydin et al., 2019) The effectiveness of fire-extinguishing balls for
combating fires was tested in controlled tests, the findings of which are presented in this article.
The goal of this continuing research effort, which included these experiments, is to create a
system of unmanned aerial vehicles (UAVs) that uses remote sensing and fire-extinguishing
balls to manage wildfires. The studies' findings suggest that, if a swarm of drones could drop
fire-extinguishing balls to optimal locations in the appropriate numbers at the right moments,
they would be a beneficial weapon to help combat wildfires.

In the work of (Soreti et al., n.d.) The primary goal of the literature study is to obtain
information about how drones are currently being used by firemen in their work. However, it took
a different approach in analyzing the fire. Since their goal was just to monitor the fire from
above. When asked about drones, all of the firefighters 17 agreed that drones provide an
informational advantage over conventional intelligence gathering methods. This research on
consumer acceptance of new technologies creates opportunities to examine in further detail
how increased preparedness and emergency preparedness can be achieved using cutting-edge
technology (drones).

In all of these UAVs, the studies confirm that further applications may be added to the
drone and that it may be used in the event of a disaster, such as a fire, that may occur in a
particular region.

2.1.2 Surveillance System

According to the findings of (Juan Jesús Roldán-Gómez1, Eduardo González-Gironda2

and Antonio Barrientos, 2021), In order to understand the primary issues firefighters, encounter

8
at work and their opinions on potential technology solutions, we conducted two surveys of
firemen. The results suggest that the most common problems are the lack of human and
material resources for all activities and the need for real-time information on the progression of
flames during extinguishing duties. When the suggested technologies assist in their work and do
not jeopardize their employment, they are favorably rated. Particularly, firemen are in favor of
the employment of drones as a tool for gathering pertinent data for fire prevention, surveillance,
and suppression operations. They think that drones can provide them with real-time information
about flames so that they can put out fires in a safer and more efficient manner. They authorize
the creation of maps that, in the circumstances of prevention and surveillance, assist coordinate
the activities for preparing vegetation and, accordingly, identify the regions with the largest risk
of fires.

In the paper of (Singh & Omkar, n.d.), The real-time Drone Surveillance System (DSS)
framework was suggested in the article, which uses drone photographs to identify one or more
people participating in violent acts. The suggested SHDL network is utilized to estimate the
posture of the individuals once the FPN network has first detected them. The SVM uses the
estimated postures to detect violent people. The suggested SHDL network accelerates training
for comparatively fewer labeled samples by combining ScatterNet features with structural priors.
Due to the high cost of collecting annotated instances, this application benefits from using fewer
labeled examples for posture estimation. In addition, the publication introduced the Aerial
Violent Individual (AVI) Dataset, which is useful for other researchers wishing to use deep
learning for applications involving aerial surveillance. On the AVI dataset, the suggested DSS
framework performs better than the cutting-edge method. This framework will be helpful in
identifying those who are involved in violent behavior in public places or at big gatherings.

2.1.3 Image and Video Processing

One of the classifications in Computer Vision that includes visual information is image or
digital processing, which is related to video processing. It is a method for processing images
and videos in order to extract and analyze valuable data for human users, which includes Object
Recognition/Detection. They allow the computer equivalent of signal conversion into digital
images, allowing the application of sight and vision to computers and allied technologies.

9
 In (Basavaprasad B., 2014), Digital image processing refers to image processing that is
relevant to a variety of key image processing applications in computer vision, such as remote
sensing, feature extraction, face identification, and picture classification. Image processing
applications include object recognition and surveillance systems, in which alive and inanimate
things may be detected and tracked using digital photos and recordings.

As described in (Anil K. Jain, 2000), Image processing and computer vision both require
pattern recognition as one of its components. It enables the development of intelligent system
computers capable of not only pattern recognition but also supervised or unsupervised
classifications.

In the paper by (Vlad Tiberiu Radu, 2019), The idea of employing drones in emergency
circumstances, particularly in firefighting activities, was widely regarded as the next high
technology instrument to be deployed in this area, with the ultimate purpose of saving lives and
money. Drones may be employed in all three stages of the disaster management cycle:
preparation, response, and recovery.

The findings show that numerous features of image processing may be combined to
produce a system that can identify a possible fire in a facility.

2.1.4 Object Detection

Object detection is a computer technique that is connected to computer vision and

image processing and is concerned with finding instances of semantic objects of a specific class
in digital pictures and videos.

According to (C. P. P., 2000), The success of these future-generation image and video
processing technologies is dependent on effective object detection algorithms. The overall
problem of object identification by computers is tough because there are a lot of factors that we
cannot account for or describe in any but the most artificial cases. The system must not draw
any conclusions about the scene illumination, the amount of items present, the size or position
of the objects, or motion, along with other things. This challenge may be approached from two
perspectives: static pictures or video sequences. If we want to find objects in static photos, we
have a straightforward pattern classification issue.

10
 In the study of (Dilip K. Prasad, 2012), In the field of computer vision, object detection is
a scientifically demanding and practically valuable topic. It is concerned with determining the
existence of different separate items in a picture. For example, significant growth has been
obtained in a controlled setting.

The detection/recognition problem has been solved in controlled environments, but it remains
unsolvable in uncontrolled environments, particularly when objects are put in random postures
in a crowded and obstructed environment.

2.2 Related Studies

2.2.1 Fire Monitoring

In the study of (Islam et al., n.d.) They released an original distributed algorithm model
that drones may use to position themselves in front of an ever-enlarging firing line. In contrast to
computationally expensive deep learning methods, the suggested method combines lightweight
image processing with fire edge recognition, which is highly desirable for drones with limited
resources. They suggested an action-reward mechanism to adjust the drones' processing rate
and speed based on the rate of shooting expansion and the available onboard computing
resources. Each drone has a finite field of vision (FoV), represented as a (wxw) square based
on the used thermal camera sensitivity and the hovering height. They found a line tangential to
the fire map by following a few steps below. The suggested three-step method offers a reliable
edge detection mechanism at a level of complexity that is much lower than the widely used
convolutional neural networks. It is especially suitable for UAVs with limited resources since it
uses a straightforward and low-complexity method to segment fire areas with great precision. In
contrast, deep-learning techniques that need higher computational resources are used. By
incorporating additional constraint criteria, they can increase segmentation accuracy.

In a study made by (Merino et al., 2011), the researchers developed a UAS-based forest
fire monitoring system. The technology incorporates the data from the fleet of various vehicles
to produce a real-time prediction of the forest fire's trajectory. The system has been put to the
test in real trials with controlled fires. These trials take into account a fleet of three aerial
systems. As they can bridge the gap between the spatial scales provided by satellite-based
systems and those based on cameras on towers, aircraft systems can be particularly useful for
fire fighting operations like fire monitoring. The UAS can modify its deployment to minimize

11
drawbacks of conventional approaches, such as smoke, or to cover locations that are more
convenient.

2.2.2 Fire Unmanned Aerial Vehicles

Moreover, according to (Yuan et al., 2017), a vision-based forest fire detection

methodology is proposed for the purpose of applying UAV-based firefighting. The proposed
method makes use of both chroma and motion features of fire in the decision-making rules to
increase the dependability and accuracy of fire detection findings. A color-based fire detection
technique with minimal computational demands was first developed with the goal of accurately
identifying suspected fire zones. The proposed motion-based fire detection system can
efficiently discriminate and track fires from background and analogues such as non-fire moving
items and disturbances owning the same color as fire, even when all objects and the backdrop
in images move due to the motion of the UAV. Real-world forest fire videos captured from
aircraft and real-time UAV footage captured in an indoor environment are used in the
experimental demonstrations. The results of the experiments have demonstrated that the
proposed method for detecting forest fires is capable of achieving good performance with
significantly increased reliability and accuracy in forest fire detection applications.

In (Pecho et al., 2019) indicates that the article's goal was to conduct research on the
application of unmanned aerial vehicles (UAVs) equipped with infrared cameras in firefighting
operations involving confined building fires. The authors conducted a fire experiment in
conjunction with numerical FDS fire simulations to test their theory. The findings of the fire test
demonstrate that the material attributes of a building's construction determine how well heat
conducts through it. The amount of time that higher back surface temperatures brought on by
fire are visible with an infrared camera is prolonged by the building envelope's high thermal
resistance. The decay phase followed by a brief period of fully developed fire (about 20 minutes)
prevented a UAV's infrared scan from detecting any temperature changes on the back surface
of the selected structures (wall and ceiling). Although, by using infrared scanning, 4 - b displays
how it is possible to view the hot air interfusion through flaws or small holes. In the absence of

12
smoke, a standard firefighting operation without the use of specialized equipment would not be
recorded.

According to (Deng et al., 2019), because of their adaptability and versatility, fire drones
are playing a more and more crucial role in emergency rescue. It is crucial to research its use in
the fields of public safety and fire protection. The use of fire-fighting drones can provide a basis
for making scientific decisions in areas like reconnaissance assistance, rescue assistance, and
monitoring and assistance. Additionally, it may improve the efficiency with which cops and staff
carry out emergency rescue operations. The operational effectiveness of firefighting drones
participating in emergency rescue missions is investigated in this study using the BPNN network
model based on the parameters of four separate elements. The results of the testing show that
the technique is effective and in line with the actual circumstances, despite the fact that the use
of fire-fighting drones is still in its early stages. Technology must still be improved and
enhanced, though. Scientific research units and firefighting units must work together to improve
the actual war experience and develop additional firefighting drone functions in order to be
ready for emergency rescue in the future.

As stated in the research of (Yang et al., 2019), In this study, a UAV-based approach for
detecting forest fires is proposed. This approach employs optical and infrared data instead of
the single source fire detection method that was previously used, it enhances the discrimination
results' robustness. The study's suggested smoke detection approach also incorporates the
frame difference method. This method only functions when the drone is hovering. The technique
doesn't work since the drone's flight causes the background to change rapidly. The high
proportion of false positive and false negative results in existing algorithms continues to be
problematic. The vast majority of algorithms are, in general, quite scenario-specific and
particularly vulnerable to outside interference. The algorithm's performance will be impacted if
the scenario gets more complicated. In this article, smoke is detected using smoke color.

Furthermore, in a study by (Remington et al., 2014), This project required expertise in

many different engineering areas to achieve. When creating the quadcopter's structural
framework, basic design principles were followed to make sure that the quadcopter wouldn't
experience any net moments, torques, or forces other than the propulsion force. An extremely
balanced and symmetrical design with the majority of the quadcopter's components placed in a
central location amid the plates of the quadcopter and the motors and propellers attached to the
ends of each evenly spaced arm came into focus as a result. According to aviation expertise,

13
opposing rotational patterns were intended to be placed next to one another in order to reduce
residual torques generated by the motors and propellers. To assist in selecting the right size of
propellers, a number of calculations were made. With knowledge of the quadcopter's weight, a
minimum thrust force may be calculated, allowing for the selection of a certain propeller size
depending on the motors employed. The development of the fire extinguishing grenade's
release mechanism accounted for the majority of the intricate design procedure. The most
efficient and reliable design for the release mechanism was found to be a four-prong claw
mechanism, and it was determined to position this release mechanism in the middle of the
quadcopter's center plates. The force applied along each of the release mechanism's prongs
was computed, and the stresses, strains, deformations, and safety factors within the PLA plastic
were then determined using a SolidWorks simulation. Although it was determined that the
material is much more resilient than necessary, PLA plastic was nevertheless chosen as the
final manufacturing material because of how simple and inexpensive it is to produce.

Lastly in research by (Sungheetha et al., 2020), Fire detection and extinguishing are
major concerns for all countries due to varied infrastructure upgrades and the implementation of
smart cities. When these catastrophes are averted globally, efficient technologies can perform
to their full potential. Early-stage fire detection is carried out utilizing IoT sensors and the WSN
platform. The World Social Network is hosted on Sensenuts platform. Using the installed sensor
nodes, real-time data gathering and analysis are carried out in either an indoor or outdoor
context. All the collected data is accommodated in cloud storage for later examination. Light
intensity, humidity, temperature, smoke, and other parameters are examined using the
ThingSpeak cloud application. Based on accurate fire detection and sensor inputs of the
environmental parameters, the real-time data is processed. This study introduces an efficient
fire detection system that integrates cloud and IoT technologies. This system makes it possible
to monitor and gather real-time data in a reasonable and cost-effective way. Integration of an
image processing strategy provides further efficiency gain. The establishment of numerous rules
increases detection accuracy. To evaluate the system's effectiveness for early-stage fire
detection, tests are run using sample images. The technique provides incredibly accurate
classification outcomes. Future work will focus on hardware optimization and sensor network
customization for fire detection in various environments.

14
 Chapter 3

METHODOLOGY

This chapter explores the approaches and techniques that the researchers will use to
carry out the study and achieve the project’s objectives.

3.1 Research Design

Research methods were the strategies, procedures, or methods used to gather

information or proof for analysis in order to learn new things or get a deeper comprehension of a
subject. There were different kinds of research techniques, each of which had its own particular
set of instruments for gathering data.

15
 Figure 3.1 Research Design

The constructive research technique is a methodical methodology that enables the

deliberate development of procedures, modules, methods, and strategies that are relevant well
beyond the case study that served as their inspiration. In general, the constructive research
approach is structured in accordance with six phases. Figure 3.1 Shows that finding and
choosing a pertinent issue with study potential is the first step. Whereas the issue is that the
firefighter lacks vision of the overall situation that is occurring on the scene. The problem's
solution and gaining background knowledge on the selected issue comes next. The researchers
will next design and build a prototype for the problem's solution. Furthermore, the researchers
will evaluate the prototype's accuracy and undertake a performance review 24 hours after it is
created. The prototype will next be evaluated based on the findings and conclusions.

3.2 Conceptual Framework

Figure 3.2 Conceptual Framework

Figure 3.2 shows how to assess the degree of fire. To get a real-time view of the
situation, the user will fly the drone into position. The captured image or video will be processed,
and the SQM of the fire will be scanned to determine what intensity of fire they would confront.

16
Using the information they obtained, the mobile application will offer excellent information and
help the situation.

3.3 Data Collection Techniques

The web-based data science environment will be the dataset utilized to train the drone's
performance. The researcher will make use of a dataset they have collected that includes
images of fire and images without fire. The data will be split into training and validation during
training. This dataset will be used to determine whether the accuracy and success rate are
sufficient for deployment. The dataset will also be used to determine if an image contains fire or
not. Images of non-fire and indoor fires collected as samples from a web-based data science
environment are displayed in Figure 3.3.1 and 3.3.2. While in figure 3.3.3 this is based on
BFP(Bureau of Fire Protection) Fire Alarm Levels. We’re going to base our data collection on
the BFP alert levels.

Figure 3.3.1

Figure 3.3.2

17
Fire Alarm Level Fire Risk Truck needed

First Alarm 2-3 Houses 4 Fire Trucks

Second Alarm 4-5 Houses 8 Fire Trucks

Third Alarm 6-7 Houses or High-Rise 12 Fire Trucks

Building

Fourth Alarm 8-9 Houses or High-Rise 16 Fire Trucks

Building

Fifth Alarm 10-11 Houses or High-Rise 20 Fire Trucks

Building

Task Force Alpha 12 X 12 Houses 24 Fire Trucks

Task Force Bravo 15 X 15 Houses 28 Fire Trucks

Task Force Charlie Affecting Significant Part of 32 Fire Trucks

Area

Task Force Delta Affecting Significant Part of 36 Fire Trucks

Area

18
 General Alarm Fire Affecting Major Part of 80 Fire Trucks
Area

Under Control Decreasing Fire Risk

Figure 3.3.3

https://thephilippinestoday.com/bfp-fire-alarm-levels/?
fbclid=IwAR3bA42PPl9oLd2fGnb4Qm41fXE9ccgCHcptqkzZXXBXujEUFcD0H9oYpAk

(BFP Fire Alarm Levels)

3.4 Block Diagram

19
 Figure 3.4 Block Diagram

The image in figure 3.4 is a system diagram where the main components or functions
are depicted by blocks connected by lines that demonstrate the relationships between the
blocks.

3.4.1 System Diagram for Mobile Application

Figure 3.4.1

Indicated in the figure above Figure 3.4.1 the system uses RaspberryPi as its
main processor. This processor will be able to process the real-time video that was taken into
frames. It will be the tool to measure the fire from above and identify the land area by square
meters of the affected areas. In doing its image processing, this will be updated and recorded in
the firebase when a fire is detected and the real-time video will be displayed through an
application in a device that is connected to the drone.

20
3.5 Hardware Design

3.5.1 Components and Materials

Figure 3.5.1

Raspberry Pi 4. The Raspberry Pi 4 is by far the most advanced version of the Pi and a
cheap computer with a number of GPIO (general purpose input/output) ports for managing
electrical devices and testing out the Internet of Things (IoT).The raspberry pi will serve as the
system's brain and evaluate the images that have been taken of the fire's range in order to
determine the fire's alert level.

21
 Figure 3.5.2

Quadcopter frame. All the other components fit into this framework (frame). It functions
as a skeleton in which several parts are arranged to evenly disperse the drone's center of
gravity. Different quadcopter frame structures with a minimum of three propeller fitting gaps are
used in different drone designs. They come in a variety of sizes and forms, and depending on
their quality, can cost anywhere from $10 to $500.

Figure 3.5.3

Motor. The rotation of the propeller requires motors. This increases the thrust needed to
move the drone. Nevertheless, the number of motors and propellers ought to match. The motors
are also constructed such that the controller can simply turn them. Their rotation improves the
drone's ability to control direction. For the drone to operate well, selecting the correct motor is
crucial. You must carefully examine a number of factors, including voltage and current, thrust
and the ratio of thrust to weight, power, efficiency, and speed, among others.

Figure 3.5.4

Electronic Speed Controller (ESC). The speed of the motor is adjusted by this
electrical control board. It serves as a dynamic brake as well. The element aids the ground pilot

22
in estimating the drone's height while it is in flight. This is achieved by calculating how much
power each motor uses individually. Power loss from the power reservoirs is related to
elevation.

Figure 3.5.5

Flight Control/Board. In the event that the drone needs to return to its takeoff site
unguided, the flight board keeps a record of the position. This feature is referred to as "return to
home." Additionally, it estimates the drone's height in relation to how much power it uses.

Figure 3.5.6

Propellers. Propellers are blades that resemble cloves and are designed to alter air
pressure. They cut through the air as they move, causing a difference in pressure at the top and
bottom of the rotors. The top side has lower pressure than the bottom, which causes the drone
to rise into the air.

23
 Figure 3.5.7

Radio Transmitter. It serves as the drone's channeled transmitter and communicator.

Each channel has a unique frequency that may direct the drone in a certain direction. Drones
need at least 4 channels to operate efficiently.

Figure 3.5.8

Battery, Electronics, and Power Distribution Cables. The drone gets its power from
the battery. Through the power distribution connections, it provides electricity to all of the
gadgets in the structure. The earliest batteries used were nickel cadmium or nickel metal hybrid
batteries. However, as lithium batteries have become more popular, their use has decreased.
They have a bigger energy storage capacity than nickel cadmium and nickel metal hybrids. A
common battery has a 3000mAh and 4V rating. An essential component for drone control and
operation is the electrical and electronic component. However, other parts can be added or
subtracted depending on the drone's intended use. Although the drone may operate without
these components, it's recommended for multitasking purposes.

Figure 3.5.9

24
 Camera. The drones have cameras on them for recording video. According on the
operator's preferences or budget, there are cameras available that can record, store, or transfer
footage. While some people use their own cameras, quality-conscious people utilize GoPro
cameras.

Figure 3.5.10

Landing gear. This structure is intended to help the drone land securely. However, it
can be waived since a skilled pilot can adjust the motor's speed for a safe landing in an
emergency. The two main categories of landing gear are. Retractable landing gear is the
alternative to fixed landing gear.

3.6 Software Design

Python. The program and language that will be used to develop, test, and debug project
code. Python will be the primary software program for the project because it has image
processing and artificial intelligence features. Python provides clear and comprehensible code,
and its accessibility enables programmers to build credible systems.

Android Studio. Offers a single platform for developing apps for Android phones.
Structured code modules enable you to separate your project into functional parts that can be
built, tested, and debugged independently. It includes performance profilers to help you better
manage your application's memory and CPU consumption, discover deallocated artifacts, find
memory leaks, enhance graphics performance, as well as analyze network requests.

Firebase. Firebase is a Google-backed mobile app platform that allows developers to

create iOS, Android, and Web apps. Firebase offers capabilities for measuring statistics,
detecting and troubleshooting app errors, as well as executing marketing and product
experiments. Using these complimentary Firebase capabilities, you can generate significant
microservices and web applications.

25
3.7 System Flowchart

The flowchart indicated the process of the system when it automatically detects a fire in
the video using image processing. After the fire is detected, the real-time video will be displayed
through a device that connects to the drone using WIFI/LTE/4G that is subsequently transmitted
to the firebase database, where it would update the user through a mobile app linked to

26
firebase. The image that was take through the video frames will be processed through
RaspberryPi. The fire will be then measured on how big the fire got based on its color detection
and displays the area in the application by square meters.

3.8 Concept of Operation

Due to the technology improvements, it is now possible to spot fires using image and
video processing. The Drone Surveillance System will be sent to the selected area where the
fire is occurring. It will survey the area to determine whether the fire is simply in the first degree
or higher before sending an alert to a mobile application. The mobile application will let the local
BFP know the present circumstances. The mobile application will show the collected image. The
user will decide what alarm level to choose and whether to SMS the fire station based on the
data obtained.

27
References:

Ausonio, E., Bagnerini, P., & Ghio, M. (2021). Drone swarms in fire suppression activities: A
conceptual framework. Drones, 5(1). https://doi.org/10.3390/drones5010017

Aydin, B., Selvi, E., Tao, J., & Starek, M. J. (2019). Use of fire-extinguishing balls for a
conceptual system of drone-assisted wildfire fighting. Drones, 3(1), 1–15.
https://doi.org/10.3390/drones3010017

Basavaprasad B, & Ravi M. (2020). A STUDY ON THE IMPORTANCE OF IMAGE

PROCESSING AND ITS APLLICATIONS. In International Journal of Research in Engineering
and Technology. https://www.researchgate.net/publication/340827009

Beijing gong ye xue yuan, Institute of Electrical and Electronics Engineers. Beijing Section, &
Institute of Electrical and Electronics Engineers. (n.d.). IEEE International Conference on
Signal, Information and Data Processing 2019 : 11-13 December, 2019 Chongqing, China.

Hariani, M. L. (2021, March 12). Determination of the Fastest Route for Fire Trucks in Cirebon
City Based on Distance, Time, Congestion and Land use.

IEEE Communications Society, & Institute of Electrical and Electronics Engineers. (n.d.). IEEE
INFOCOM 2019 - IEEE Conference on Computer Communications Workshops (INFOCOM
WKSHPS).

IEEE Control Systems Society, IEEE Robotics and Automation Society, & Institute of Electrical
and Electronics Engineers. (n.d.). 2017 International Conference on Unmanned Aircraft
Systems (ICUAS’17) : final program and book of abstracts : June 13-16, 2017, Miami Marriott
Biscayne Bay, Miami, FL 33132.

Islam, S., Huang, Q., Afghah, F., Fule, P., & Razi, A. (n.d.). Fire Frontline Monitoring by
Enabling UAV-Based Virtual Reality with Adaptive Imaging Rate.

28
Khan, I., Singh, S. K., Ali, I., Scholar, Mt., & Associate Professor, J. (2016). Fire Detection Using
Image Processing Based onColorAnalysis. International Research Journal of Engineering and
Technology. www.irjet.net

Marquez Consuelo “Fire hits houses in Mandaluyong City | Inquirer News” Inquirer.Net (2019)
https://newsinfo.inquirer.net/1196019/fire-breaks-out-in-mandaluyong-city-residential-area

Michael Lim “Fire hits Barangay Addition Hills” Metro News Central (2019)
https://metronewscentral.net/mandaluyong/barangay-front/metro-cities/fire-hits-barangay-addition-hills

Nazar Zadeh, N. R., Abdulwakil, A. H., Amar, M. J. R., Durante, B., & Reblando Santos, C. V. N.
(2021). Fire-fighting UAV with shooting mechanism of fire extinguishing ball for smart city.
Indonesian Journal of Electrical Engineering and Computer Science, 22(3), 1320–1326.

Peter M Ngau, Sharon J Boit “ Community fire response in Nairobi's informal settlements” Sage
Journals (2020) https://journals.sagepub.com/doi/full/10.1177/0956247820924939

Singh, A., & Omkar, S. N. (n.d.). Eye in the Sky: Real-time Drone Surveillance System (DSS)
for Violent Individuals Identification using ScatterNet Hybrid Deep Learning Network.

Soreti, C., Radu, V. T., Schmidt, A., & Mehmood, K. S. (n.d.). Use of drones for firefighting
operations Title: Use of drones for firefighting.

Sungheetha, Dr. A., & Sharma R, Dr. R. (2020). Real Time Monitoring and Fire Detection using
Internet of Things and Cloud based Drones. Journal of Soft Computing Paradigm, 2(3), 168–
174. https://doi.org/10.36548/jscp.2020.04

Tupas Emmanuele “Mandaluyong fire leaves 72 families homeless” PhilStar (2022)

https://www.philstar.com/nation/2022/06/01/2185114/mandaluyong-fire-leaves-72-families-
homeless

Wang, K., Yuan, Y., Chen, M., Lou, Z., Zhu, Z., & Li, R. (2022). A Study of Fire Drone
Extinguishing System in High-Rise Buildings. Fire, 5(3), 75. https://doi.org/10.3390/fire5030075

29
30