VISVESVARAYA TECHNOLOGICAL UNIVERSITY

“Jnana Sangama”, Belagavi-590014, Karnataka, India

A PROJECT REPORT ON

“VISION BASED FIRE DETECTION”

Submitted in partial fulfilment of the requirements for the award of

BACHELOR OF ENGINEERING
IN
COMPUTER SCIENCE AND ENGINEERING

For the Academic year

2020 – 2021

Submitted by
SURAKSHA N 4SH17CS103
SUSHMITHA 4SH17CS105
VANISHRI RAMA MOGER 4SH17CS109
VASTHAVI 4SH17CS110

Under the guidance of:

Mrs. Nisha V C

DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

Shree Devi Institute of Technology
Kenjar, Mangaluru-574142
 SHREE DEVI INSTITUTE OF TECHNOLOGY
KENJAR, MANGALURU- 574142
DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

CERTIFICATE

This is to certify that the Project work entitles “VISION BASED FIRE DETECTION ” is
a bonafide work carried out by SURAKSHA N, SUSHMITHA, VANISHRI RAMA
MOGER, VASTHAVI bearing USN 4SH17CS103, 4SH17CS105,
4SH17CS109,4SH17CS110 in fulfillment for the 8th semester Project of Bachelor of
Engineering in Computer Science and Engineering of Visvesvaraya Technological
University, Belagavi during the year 2020-2021. It is certified that all correction/
suggestions indicated for Internal Assessment have been incorporated in the report. The
report has been approved as it satisfies the academic requirement in respect of the Technical
Seminar work prescribed for the said degree of Bachelor of Engineering.

Signature of the Guide Signature of the HOD Signature of the Principal

Prof. Nisha V C Prof. Anand S Uppar Dr. K E Prakash

EXTERNAL VIVA

Name of the Examiners Signature with date

1. ----------------------------------- --------------------------
2. ----------------------------------- --------------------------
 SHREE DEVI INSTITUTE OF TECHNOLOGY
KENJAR, MANGALURU – 574142

Department of Computer Science and Engineering

DECLARATION

We SURAKSHA N, SUSHMITHA, VANISHRI RAMA MOGER, VASTHAVI bearing

USN 4SH17CS103, 4SH17CS105, 4SH17CS109,4SH17CS110 students of Eighth semester
Bachelor of Engineering, Computer Science and Engineering, Shree Devi Institute of
Technology, Mangaluru declare that the Project work entitled “VISION BASED FIRE
DETECTION” has been duly executed by us under the guidance of Mrs. Nisha V C, Asst.
Professor, Department of Computer Science and Engineering. Shree Devi Institute of
Technology, Mangaluru and submitted the 8th semester Project Work for partial fulfillment
of the requirements for the award of degree of Bachelor of Engineering in Computer
Science Engineering during the year 2020-2021.

Date: 17/07/2021 SURAKSHA N

Place: Mangaluru SUSHMITHA

VANISHRI RAMA MOGER

VASTHAVI
 ACKNOWLEDGEMENT

A successful project is a fruitful culmination of the efforts of many people. Some

directly involved and others who have quietly encouraged and extended their invaluable
support throughout its progress.
We would like to convey my heartfelt thanks to our Management for providing the
good infrastructure, laboratory facility, qualified and inspiring staff whose guidance was of
great help in successful completion of this project.
We would like to express a warm thanks to Dr. K E Prakash, Director, Shree Devi
Institute of Technology, Kenjar for providing the necessary facilities to successfully carry
out this project.
We are extremely delightful and thankful to our beloved principal Dr. K E Prakash,
Shree Devi Institute of Technology, Kenjar for providing the congenial atmosphere and
necessary facilities for achieving the cherished goal.
With heartiest gratitude, We would like to thank Prof. Anand S Uppar, HOD,
Department of Computer Science and Engineering for his support, guidance and
encouragement.
We are profoundly indebted to Mrs. Nisha V C Guide, Assistant Professor,
Department of Computer Science and Engineering, for their guidance throughout the work
by innumerable acts of timely advice and encouragement.
We also thank all other teaching staff and non-teaching staff for allowing us to carry
out the project work.
Finally we would like to thank our family for their support and understanding, to
whom we owe so much.

SURAKSHA N

SUSHMITHA

VANISHRI RAMA MOGER

VASTHAVI
 ABSTRACT

Fire is very dangerous that brings great loss of life and properties It’s not possible for human
being to continuously monitor the accidental fire situations. Therefore in such cases fire
detection using ML came into picture. Fire detection system is based on detection of fire
from video acquisition input data. This is done with the help of digital image processing
techniques. It is based on vision based fire detection system. This video based fire detection
systems has stages such as frame conversion, fire color detection, alert message and sound
and the count of fire occurrence in a region. Fire alarm will be activated when the system
detects the occurrence of fire in a certain position.
 CONTENTS

CHAPTERS PAGE NUMBER

1. INTRODUCTION 1

1.1 Project Introduction

1.2 Problem Description

2. LITERATURE REVIEW 2-5

2.1 General Introduction

2.2 Literature survey

2.3 Summary

3. PROBLEM FORMULATION 6

3.1 General

3.2 Problem Statement

3.3 Objectives

3.4 Summary

4. REQUIREMENTS AND METHODOLOGY 7-8

4.1 Software Requirements

4.2 Hardware Requirements

4.3 Methodology

5. SYSTEM DESIGN 9-11

6. IMPLEMENTATION 12-16

7. SYSTEM TESTING,RESULTS 17-22

8. CONCLUSION AND SCOPE 23-24

9. REFERENCES 25
 LIST OF FIGURES

FIGURE CONTENT PAGE NUMBER

5.1 Design of the proposed fire detection using ML 9

5.2 Flowchart of the proposed System 10

7.2.1 Home Page of the Vision based fire detection system 18

7.2.2 Login Page of the Vision based fire detection system 18

7.2.3 Register Page of the Vision based fire detection 19

system

7.2.4 Input Page of the Vision based fire detection system 19

7.2.5 Detection of the fire from the input 20

7.2.6 Final Result Page of the Vision based fire detection 20

system

LIST OF TABLES

TABLE NUMBER CONTENT PAGE NUMBER

4.1 HARDWARE REQUIREMENT 7

4.2 SOFTWARE REQUIREMENT 7

7.1 TEST CASES OF VALIDITY 16

LIST OF ACRONYMS AND ABBREVIATIONS

SVM Support Vector Machine

KNN K Nearest Neighbour

RAM Random Access Memory

HSV Hue Saturation and Value

PCA Principal component analysis

HTML Hypertext mark-up language

OpenCV Open Source Computer Vision

Software Library

csv comma-separated value

VISION BASED FIRE DETECTION 2020-2021

Chapter 1

INTRODUCTION

1.1 Project Introduction

Fire detection has become important for both personal and commercial applications. The
delay in the detection of fire in fire detection systems continue to be a life-threatening
problem for living things. The traditional fire detection system is based on sensors that rely
on heat, pressure, smoke and temperature conditions works slower and hence time
consuming. Video based fire detection system can be reliable alternative to the existing
traditional fire detectors.

1.2 Problem Description

Fire is one of the most common and dangerous disasters and usually leads to environmental
damage, large economic losses and many casualties. Fire detection has become important
for both personal security and commercial applications.

Traditional fire detection system based on sensors works slower and hence time consuming.
Sensor based fire detection requires a greater number of sensor to cover a large area and also
waits for some condition to trigger the sensors. Besides, it takes relatively long time for
particles to reach the sensor and activate them.

Dept. of CSE, SDIT Page 1

VISION BASED FIRE DETECTION 2020-2021

Chapter 2

LITERATURE REVIEW

2.1 General Introduction

Literature Survey is an important activity, which we have to do while gathering information
about a particular topic. It will help us to get required information or ideas to do work. The
following paragraphs discuss the related work and issues in the area Prediction and Analysis
of Energy Consumption in a Computer Laboratory using machine learning algorithm.

2.2 Literature Survey

Sreesruthi Ramasubramanian et ed. [2] in “Fire Detection using Artificial Intelligence for Fire-
Fighting Robots” have told that Fire-fighting robots are used in indoor environments to detect
fires and extinguish them. Sensors such as flame sensors are currently used to detect fire in
firefighting robots. The disadvantage of using sensors is that fire beyond a threshold distance
cannot be detected. Using artificial intelligence techniques, fire can be detected in a wider
range. Haar Cascade Classifier is a machine-learning algorithm that was initially used for
object detection. The results obtained using Haar Cascade Classifier were not very accurate,
especially when multiple fires had to be detected.

Abolfazl Zargari Khuzani, Rakshit Agrawal, Najmeh Mashhadi [2], proposed a real-time
fire detection system used to identify fire between fire and non-fire cases in video sequences.
The proposed method consisted of four steps: (1) Dividing video sequences into three-
second video clips. (2) Applying a color space model to segment the fire and fire-like regions
of each frame. (3) Employing a feature extraction scheme to create a feature pool of spatial
and frequency domain, considering motion and time variations. (4) Reducing the size of the
feature pool by Kernel-PCA. (5) Training a multi-layer neural network to classify 502 video
clips into the fire and non-fire categories. The results showed that the proposed framework
can provide a reliable prediction with a low false rate.

Dept. of CSE, SDIT Page 2

VISION BASED FIRE DETECTION 2020-2021

Chuanchen Li, Yong Bai[3], propose two approaches to fasten the learning process and
improve the detection performance based on transfer learning . Firstly propose an approach
by using and fine-tuning the existing transfer learning models. The other proposed approach
is to use multiple transfer learning models to extract the image features, then fuse these
features, and lastly employ the machine learning classifier for flame detection. In the latter
proposed approach, ensemble learning method is also investigated to enhance the classifying
performance. The experimental results demonstrate that both these two approaches can
achieve very high detection accuracy for fire flame image detection.

Jubeena. B. Maheen, Aneesh R P[4], Camera surveillance system is a predominant part in

building automation system and thus computer vision based fire detection system are being
developed to embed with camera. The identification of blaze edges defines a practice of
determining margin between the region of thermal response or not. Novel approach is
proposed to distinguish the fire region from the video frame using color correlogram feature.
Saliency based features are also extracted from the segment. The color probability of
neighboring pixel is estimated with color auto correlogram feature. Machine learning
algorithm is implemented to train the feature vector. With the naive bayes classifier, the
proposed system could test Fire 2018 dataset and obtained 96.97 % accuracy.

Jareerat Seebamrungsat, Suphachai Praising, and Panomkhawn Riyamongkol[5], A fire

detectionsystem based on light detection and analysis is proposed in the paper. This system
uses HSV and YCbCr color models with given conditions to separate orange, yellow, and
high brightness light from background and ambient light. Fire growth is analysed and
calculated based on frame differences. The HSV color model is used to detect information
related to color and brightness. The YCbCr color model is used to detect information related
to brightness because it can distinguish bright images more efficiently than other color
models. The overall accuracy from the experiments has been greater than 90%.

Xuehui Wu,,Xiaobo Lu, Henry Leung[6], —proposes a novel method for fire and smoke
detection using video images. The ViBe method is used to extract a background from the
whole video and to update the exact motion areas using frame-by-frame differences.
Dynamic and static features extraction are combined to recognize the fire and smoke areas.
For static features, deep learning is used to detect most of fire and smoke areas based on a
Caffemodel. Another static feature is the degree of irregularity of fire and smoke. An

Dept. of CSE, SDIT Page 3

VISION BASED FIRE DETECTION 2020-2021

adaptive weighted direction algorithm is further introduced. To further reduce the false
alarm rate and locate the original fire position, every frame image of video is divided into
16×16 grids and the times of smoke and fire occurrences of each part is recorded. All clues
are combined to reach a final detection result. Experimental results show that the proposed
method in this paper can efficiently detect fire and smoke and reduce the loss and false
detection rates.
Ke Chen, Yanying Cheng, Hui Bai, Chunjie Mou, Yuchun Zhang[7], an image fire detection
algorithm based on support vector machine is proposed by studying the features of fire in
digital image. Firstly, the motion region is extracted by the inter-frame difference method
and regarded as the Suspected fire area. Then, the uniform size is sampled again. Finally,
the flame color moment feature and texture feature are extracted and input into the support
vector machine for classification and recognition. Data sets were formed by collecting
Internet resources and fire videos taken by oneself and the trained support vector machine
was tested. The test results showed that the algorithm can detect early fire more accurately.

S.R.Vijayalakshmi, S.Muruganand[8], Fire alarm system is based on detection of fire from

video acquisition input data. This is done with the help of digital image processing
techniques and embedded vision. It is based on vision based fire detection system. This
approach integrates colour, spatial, temporal and motion information to locate fire regions
in video frames. The characteristics of burning flame and spatial temporal features of smoke
combination is used to remove spurious fake fire regions. This video based fire alarm
systems has stages such as frame conversion, fire and smoke colour detection, motion
detection, spatial temporal analysis. Fire alarm will be activated when the system detects the
occurrence of fire in a certain position and time interval surpasses the threshold.

Dept. of CSE, SDIT Page 4

VISION BASED FIRE DETECTION 2020-2021

2.3 Summary

Important aspects we found in our reading can be described as follows: The fire detection is
done by image processing and then applying machine learning algorithm to classify whether
it is fire or not. The image of the fire when processed color model is applied to segment the
fire like regions of the frame. Color model such as HSV, YCbCR is applied andNovel
approach is proposed to distinguish the fire region from the video frame using color
correlogram feature. The ViBe method is used to extract a background from the whole video
and to update the exact motion areas using frame-by-frame differences. Finally, the flame
color moment feature and texture feature are extracted and input into the machine algorithm
for classification and recognition.

Dept. of CSE, SDIT Page 5

VISION BASED FIRE DETECTION 2020-2021

Chapter 3

PROBLEM FORMULATION
3.1 General
Before attempting to solve a problem, we need to first formulate or define the problem. It is
important to precisely define the problem you intend to solve. Problem formulation is the
act of a problem, determining the cause of the problem and, identifying the solution.

3.2 Problem Statement

Fire detection system detects the occurrence of fire and alerts the respective department by
raising the alarm. The fire detectors which work based on conventional sensors such as
infrared sensors, optical sensors and ion sensors cannot be used in outdoor or large areas
and cannot locate the fire regions. It is usually time consuming. In order to facilitate earlier
detection of fire, a vision-based fire detection system based on machine learning algorithms
is developed.

3.3 Objectives of the Present Study

The objectives of the proposed project are as follows:

To detect fire while they are still small and have not grown too large.

1. To develop a low cost fire detection system in real time.

2. Detect the fire in video frame.

3. Send alert message.

3.4 Summary
The traditional fire detection system are widely used, but they have lots of weaknesses in
practice. Sensor based fire detection detectors cannot be used in outdoor and large areas.
They are time consuming and wait for some condition to trigger the sensors. Therefore a

Dept. of CSE, SDIT Page 6

VISION BASED FIRE DETECTION 2020-2021

video based fire detection system can be reliable alternative to the existing traditional fire
detection system cannot be used. It creates a higher possibility of fire detection at an early
stage.

Chapter 4

Requirements and Methodology

4.1 Requirements
The proposed project consists of following modules:
4.1.1 Hardware requirements
4.1.2 Software requirements

4.1.1 Hardware Requirements

The hardware requirements for the proposed project are depicted in the table below: Table
4.1: Hardware requirements
Sl. No Hardware/Equipment Specification

1. Processor Intel i3 Core Processor

2. Clock speed 2.20GHz

3. Monitor 1024*768 Resolution,

Colour

4. RAM 8GB

4.1.2 Software Requirements

The software requirements for the proposed project are depicted in the table below:

Table 4.2: Software requirements

Sl. No Software Specification

1. Operating system Windows 10

2. Languages HTML, python

3. Tools used PyCharm, OpenCV

Dept. of CSE, SDIT Page 7

VISION BASED FIRE DETECTION 2020-2021

4.2 Methodology Used Steps followed for implementing the

Proposed Project
Steps followed while implementing fire detection using machine learning is that Initially the
video file is given as training dataset.

• The given video file is converted into frames.

• The attributes of fire such as area, perimeter, major axis, minor axis, entropy, color
is extracted from the image.

• These attributes are stored in csv file which serves as training dataset.

• Training dataset now contains file with fire and without fire.

• When a real time video is given as input which serves as testing dataset.

• The testing data is compared with training data.

• The testing data is classified into a video with fire and without fire using KNN and
SVM algorithm.

• If the fire is detected an alert sound is played and a warning message is sent to the
fire department.

Dept. of CSE, SDIT Page 8

VISION BASED FIRE DETECTION 2020-2021

Chapter 5
SYSTEM DESIGN
5.1 System Design
System design is a one important phase in software or system development. System design
can be defined as method of defining different modules required for software or system to
fulfil all requirements.

5.1.1 Architecture of proposed system

System design shows the overall design of the proposed Vision Based Fire Detection, which

Figure 5.1.1: Design of the proposed Vision Based Fire Detection using ML

5.1.1.1 Phases in Vision based Fire Detection system.

Dept. of CSE, SDIT Page 9

VISION BASED FIRE DETECTION 2020-2021

The Vision based fire detection system based on the algorithm SVM and KNN proposed
based on the establishment of positive and negative fire sample datasets. The image from
the video clip is segmented which completely separates the foreground from the
background. Then it is compared with features of the fire which is stored as attributes in the
csv file. If the attributes are matched then the alert message is activated and a warning
message is sent to the respective email address. Later the occurrence of the fire in a particular
region is analysed and the fire prone region is detected.

5.1.1.2 System Flowchart

The flowchart of the proposed system is depicted in figure 5.2 shown below:

Figure 5.1.1.2: Flowchart of the proposed system

Dept. of CSE, SDIT Page 10

VISION BASED FIRE DETECTION 2020-2021

Chapter 6

Implementation

The proposed project consists two implementation modules.

6.1 Pseudocode
Pseudocode for the SVM algorithm used.
Define number of features+1 as F and SVs+1 as SV
FOR each SV
FOR each feature of the SV
Read streamed data
Convert it to float
Store into array_SVs[SV][F]
END FOR
END FOR
Read streamed data
Convert it to float
Store into array_ay[0](b value)
FOR each SV
Read streamed data
Convert it to float
Store it into array_ay[SV]
END FOR
FOR each feature
Read streamed data
Convert it to float
Store into array_test[F]
END FOR
FOR each feature
Clear array_AC[F]
END FOR
FOR each SV

Dept. of CSE, SDIT Page 11

VISION BASED FIRE DETECTION 2020-2021

FOR each feature of the SV

array_AC[F]+=array_ay[SV]*array_SVs[SV][F]
END FOR
END FOR
FOR each feature
Distance_value+=array_AC[F]*array_test[F]
END FOR
Distance_value-=b
If(Distance_value>=th)THEN
RETURN 1
ELSE
RETURN-1
END IF

Pseudocode for the KNN algorithm used.

k-Nearest Neighbour Classify(X,Y,x)//Training data, Y:class labels of X,
x:unknown sample for i=1To n

do

Compute distance d(X,x) end for

Compute set I containing indices for the k smallest istances d(X,x)
Return majority label for {Yi where I belongs to I)

Dept. of CSE, SDIT Page 12

VISION BASED FIRE DETECTION 2020-2021

6.2 Implementation Code

import csv import math
import operator

def euclideanDistance(instance1, instance2, length):

distance = 0

for x in range(length):

distance += (pow((float(instance1[x]) - float(instance2[x])), 2))

return math.sqrt(distance)

def getNeighbors(trainingSet, testInstance, k):

distances = [] length = len(testInstance)-1

for x in range(len(trainingSet)):
dist = euclideanDistance(testInstance, trainingSet[x], length)
distances.append((trainingSet[x],dist))
distances.sort(key=operator.itemgetter(1))neighbors=[]

for x in range(k):

neighbors.append(distances[x][0]) return neighbors

def getResponse(neighbors): classVotes = {}

for x in

range(len(neighbors)):
response = neighbors[x][-1]

if

response in classVotes:
classVotes[response] += 1

else:

Dept. of CSE, SDIT Page 13

VISION BASED FIRE DETECTION 2020-2021

classVotes[response=1sortedVotes=sorted(classVotes.items(), key=operator.itemgetter(1),
reverse=True)

return sortedVotes[0][0]

def getAccuracy(testSet, predictions):

correct = 0 for x in range(len(testSet)):

if

testSet[x][1] == predictions[x]:

correct += 1

return (correct / float(len(testSet))) * 100.0

trainingSet = [] testSet = [] with

open('data/train.csv', 'r')

as csvfile:

lines = csv.reader(csvfile)

dataset = list(lines)

# print(dataset)

for x in range(len(dataset) - 1):

for y in range(5):

dataset[x][y] = float(dataset[x][y])
trainingSet.append(dataset[x])

with open('data/test.csv', 'r') as csvfile1:

lines1 = csv.reader(csvfile1)

# print(lines1) dataset1 = list(lines1)

# print(dataset1)

Dept. of CSE, SDIT Page 14

VISION BASED FIRE DETECTION 2020-2021

for p in range(len(dataset1)):
for q in range(5):
dataset[p][q] = float(dataset[p][q])

testSet.append(dataset1[p])

print("trainingset:", trainingSet)
print("testingset:",testSet)

# print("1:",len(trainingSet))

# print("2:",len(testSet)) k = 1 predictions = []

for

x in range(len(testSet)):

neighbors = getNeighbors(trainingSet, testSet[x], k)

response = getResponse(neighbors)

print("\nNeighbors:", neighbors) print('\nResponse:',response)

predictions.append(response)

accuracy = getAccuracy(testSet, predictions) print('Accuracy: ' + repr(accuracy) + '%')

import matplotlib.pyplot as plt

x = [0, 1, 2]

y =[accuracy,0, 0]
plt.title('Accuracy')
plt.bar(x, y) plt.show()

Dept. of CSE, SDIT Page 15

VISION BASED FIRE DETECTION 2020-2021

Chapter 7

System Testing, Results and Discussion

7.1 System Testing

The purpose of testing is to discover errors. Testing is the process of trying to discover every
conceivable fault or weakness in a work product. It provides a way to check the functionality
of components, sub-assemblies, assemblies, and/or a finished product. It is the process f
exercising software with the intent of ensuring that the software system meets its
requirements and user expectations and does not fail unacceptably.

Software testing is the process of checking whether the developed system is working
according to the original objectives and requirements. The software testing process
commences once the program is created and the documentation and related data structures
are designed. Software testing is essential for correcting errors. Otherwise, the project is
not aid to be complete.

Table 7.1.1: Test cases for the validity

TEST STAGE INPUT EXPECTED OBTAINED REMARK
CASE SPECIFICATION OUTPUT RESULT
NO.

1 Login page Check user login User should User login Pass
with expected data. Login successfully

2 Register Required details of User should User Pass

page. the user are given register registered
as input. successfully

Dept. of CSE, SDIT Page 16

VISION BASED FIRE DETECTION 2020-2021

3 Image Select the mp4 Should select Noise is Pass

acquisition video containing image from reduced.
and reducing fire the video
noise and reducing
the noise.

4 Alert sound After noise Should Alert sound is Pass

generation reduction video generate alert generated
File is given as sound when
input fire is detected
5 Alert After noise Should create Alert message Pass
message reduction video file an alert is created and
generation is given as input message sent to
when fire is respective
detected authorities

In table 7.1 test cases for the validity given to the system are described. In the Vision based
fire detection system admin should login with the valid mail, password and press login
button. Then the video is given as input which may or may not contain fire. After getting
video image acquisition is done which divides into frames and reduction of noise is done
then certain feature of fire is extracted. Later it generates an alert sound and also the message
is generated saying fire is detected to respective authorities. After these when fire is detected,
in a database occurrence of fire is recorded.

Dept. of CSE, SDIT Page 17

VISION BASED FIRE DETECTION 2020-2021

7.2 Result Analysis

7.2.1 Home Page

This is the home page of the Fire detection system where the user can login or register to the
system.

Figure 7.2.1: Home page of the Vision based fire detection system.

7.2.2 Login page

This is the login page of the system where an already existing user can login to the system by
entering his username and password.

Figure 7.2.2: Login page of the Vision based fire detection system.

Dept. of CSE, SDIT Page 18

VISION BASED FIRE DETECTION 2020-2021

7.2.3 Register page

This is the register page where a new user can register into the system by entering the required
details.

Figure 7.2.3: Register page of the Vision based fire detection system.

7.2.4 Input page.

This is the input page of the which provides the user to select whether the input is given
through real time camera or from a video file.

Figure 7.2.4: Input page of the Vision based fire detection system.

Dept. of CSE, SDIT Page 19

VISION BASED FIRE DETECTION 2020-2021

7.2.5 Detection of Fire

This figure describes the detection of the fire from the given input.

Figure 7.2.5: Detection of the fire from the input.

7.2.6 Final Outcome page.

This page displays the final outcome where in if the fire is detected it displays and sends an
alert message to the fire department. It also displays the fire prone region.

Figure 7.2.6: Final result page of the Vision based fire detection system.

Dept. of CSE, SDIT Page 20

VISION BASED FIRE DETECTION 2020-2021

7.3 Summary
The vision based fire detection system has a home page where an already existing user can
login to the system. If he is a new user he can register to the system by entering the required
details. The input page allows the user to select the way in which input is fed to the system.
The system takes the input and processes the input file to detect the fire. If the fire is detected
in the input file then the system displays the message that the fire is detected. The alert sound
is played and a warning message is sent to the respective person or the fire department. It
also displays the region which is prone to the fire accident.

Dept. of CSE, SDIT Page 21

VISION BASED FIRE DETECTION 2020-2021

Chapter 8

Conclusion and Scope for Future Work

8.1 Conclusion
Fire detection systems are vital in saving people’s lives, and preventing hazards before they
get out of control. We proposed an system for fire detection in video sequence . Our system
takes the image and represents the extraction of edge features from image. From the obtained
features it is compared with training dataset and classifier such as SVM and knn is applied
which informs the system whether the image is fire or without fire. The results confirmed
that proposed method provides reliable detection performance.

8.2 Scope for Future Work

The proposed system can be further improved by collecting data and applying other machine
learning algorithms to increase the accuracy of the system to reduce the false positive. The
detection methods can be optimised as per the nature of area covered by the detection system
and also different techniques for the detection can be combined together to reduce the false
alarm rate and improve the response time.

Dept. of CSE, SDIT Page 22

VISION BASED FIRE DETECTION 2020-2021

REFERENCES

[1] Ke Chen, Yanying Cheng, Hui Bai, Chunjie Mou, Yuchun Zhang, “Research on
Image Fire Detection Based on Support Vector Machine” 2019 IEEE.

[2] Sreesruthi Ramasubramanian school of Engineering and Physical Science, “Research

on Fire Detection using artificial Intilligence for firefighting Robots”,2020 IEEE

[3] Yeung Lee dept of Multimedia Engineering, “Data Labelling Research for Deep
Learning Based Fire Detection” 2019 IEEE

[4] Jubeena B Maheen Kerala University, ‘Machine Learning Algorithm for Fire
Detection using Color Correlogram, 2019 @nd international Conference on
Intelligent computing, ICICICT

[5] Ngoanidzashe A Mwedi; Nnamdi I, Nwulu and Saheed lekan Gbadamosi, “Machine
learning Application for Fire detection in a residential Building, 2019 6Th IEEE
International Conference on Engineering technologies and Applied Sciences.

[6] A Shreyl Oliver Assocaite Professor, St Joseph College of Engineering Chennai,

“Detection of Forest Fire using Convolution Neutral networks, IEEE 7th International
Coference on Smart Structures and System 2020.

[7] Guohua wang South Canara Agricultural university “Research based on forest smoke
detection based on deep learing and background modelling, IEEE 2020

[8] Xuehui Wu 1,2 , “An adaptive Threshold Deep Learing Method for Fire And smoke
detection, 2017 IEEE conference on Systems.

Dept. of CSE, SDIT Page 23

VISION BASED FIRE DETECTION 2020-2021

Dept. of CSE, SDIT Page 24