© 2022 IJRAR April 2022, Volume 9, Issue 2 www.ijrar.

org (E-ISSN 2348-1269, P- ISSN 2349-5138)

Smart CCTV Surveillance System

Dr. Abhijit Patankar1, Yash Jadhav2, Deepkumar Brahmbhatt3, Ashutosh Raut4, RitujaChavan5

1,
Assosicate Professor, Information Technology, DY Patil College of Engineering, Akurdi-Pune 411044.
2,3,4,5
BE, Information technology, Dr.D.Y. Patil college of engineering, Akurdi - Pune.411044.

ABSTRACT

The venture Smart CCTV camera reconnaissance framework is to upgrade the CCTV camera-based security frameworks, which by
and by exist in better places. The Project Security System by utilizing CCTV Camera is planned utilizing remote innovation. The
utilization of an observation framework for is turning out to be more significant now a days in lodging purposes and Home security.
We will add various capacities like Fire location, face acknowledgment.

Keywords: CNN,ML,Dataset,Training Module.

I. INTRODUCTION
The venture Smart CCTV camera reconnaissance framework is to upgrade the CCTV camera-based security frameworks, which
by and by exist in better places. The Project Security System by utilizing CCTV Camera is planned utilizing remote innovation. The
utilization of an observation framework for is turning out to be more significant now a days in lodging purposes and home security.
We will add various capacities like Fire location, face acknowledgment Fire is one of the major causes of financial losses and
humanistic disaster in the world, which can occur in various places such as schools, offices, companies, residences, mines and
forests. In order to minimize the damage caused by fire and reduce the financial losses attached, it is important to prevent the spread
of fire in time. Fire detection is an important key to the area surveillance system as a timely warning to report a fire start [1]. Most
fire detectors use pointers to measure fire potential in the immediate vicinity. In order to minimize the false alarm detection of the
fire detection system, the detector needs to separate the smoke from the fire hazard because activating a false fire alarm can be
very costly and can disrupt production and equipment. Therefore, in order to overcome the interruptions connected to the false
alarm from the fire detection system, it is very important to check the current technology in fire detection, in order to provide a
reliable and efficient system. The development of deep learning, computer vision and digital camera technology has aided the use
of intelligent video fire detection approach, thereby upgrading the fire detection systems to an intelligent and vision-based system.
Reference presents an experimental analysis using neural network approach for detecting and analysing the fire and false alarm
conditions. More recently, some researchers have been working to detect the effects of fire on a computer screen. They focus on
fire features and color models; this method uses cameras, smart devices, wireless networks and image processing techniques. This
is a model and used to detect local fire incidents. Similarly, classification approach based on machine learning have been employed
for fire detection problems in , contours and bounding boxes are used to indicate the fire pixels detection on an image. This
mechanism distinguished if there exist more than a single fire in the image. Reference develops an algorithm for extracting the data
characteristics of fire such as smoke density, temperature and carbon monoxide density using a fuzzy system. Reference proposes
an artificial intelligence algorithm for fire detection using shape characteristics of fire from video footages. This paper presents a
novel intelligent fire detection system method. A VGGNet architecture was derived from CNN network with the aim of achieving a
good image classification performance, minimal false alarm and high fire detection rate.

IJRAR1CNP017 International Journal of Research and Analytical Reviews (IJRAR) www.ijrar.org 105
 © 2022 IJRAR April 2022, Volume 9, Issue 2 www.ijrar.org (E-ISSN 2348-1269, P- ISSN 2349-5138)
II. Literature survey

There have been many researches over the years that focus on the salient features of the images. Borges used additional features such as
surface, area, and boundary of the fire area to detect fires. , authors used Convolution Neural Networks (CNN) for the detection of fire.
In , authors explained that CNN is deep learning algorithm which takes images as input and allocate weights in order to
differentiate from others. CNN formation is similar to the formation of neurons in the human brain. CNN compares images into
episodes. The pieces you look at are called features. Convolutional- layer creates a line of image and image by multiplying the
pixel of each image with the corresponding element pixel. After assembling, divide by the total number of pixels on the element.
The Rectified Linear Unit (ReLU) removes all negative values from the filtered image and replaces it with zero. ReLU is used to
prevent values from summarizing to zero. Blending layer reduces image stack into smaller size with the help of window size and step
size. The composite layer goes through the window in the filtered image. The final output needs to be flattened before it can be
delivered to the normal neural to separate. The output of flatten layer is fed to a feed forward and backpropagation employed to
every repetition of the training . In [7], early detection of disaster events is achieved with the help of computer computer
infrastructure. The computer structure of the fog provides a prediction based on machine learning with weather sensors. Fog
computing and Convolutional Neural Networks (CNN) have merged with Unmanned Aerial Vehicles (UAV) to detect forest fires
early.Even CNN provides excellent performance, it is very difficult to catch dynamic behavior of fire, which can be improved by
using RNN (recursive-type neural networks). RNN model accumulates the features for decision making through the memory cell
which preserve the recurrent behavior and internal states. However, repetition of recursions is usually limited, which makes it hard
to capture long-term behaviour to make a decision.

Sr no Title Author Year Method

1 Design of a smart fire KB Deve GP 2016 IOT
detection system Hancke; BJ Silva
2 Design and Oh-Hyun Kwon 2008 IOT
Implementation of Department of
Fire Detection System Computer Engineering,
Tongmyong University,
South
Korea
; Sung-Min Cho; Sun-
Myung Hwang

3. IoT-Based Fire Asma Mahgoub; 2019 IOT

Alarm System Nourhan Tarrad;
Rana Elsherif;
Abdulla Al-Ali;
Loay Ismail
4 About automatic fire Huide Liu; Suwei Li; 2010 IOT
alarm systems Lili Gao; Tao
research Wu
5 Computer Vision and Nahid Chowdhury; 2019 Computer Vision
Smoke Sensor Dewan Ruhul
Based Fire Mushfiq; AZM
Detection System Ehtesham Chowdhury

IJRAR1CNP017 International Journal of Research and Analytical Reviews (IJRAR) www.ijrar.org 106
 © 2022 IJRAR April 2022, Volume 9, Issue 2 www.ijrar.org (E-ISSN 2348-1269, P- ISSN 2349-5138)
III. PROPOSED SYSTEM

Fig 1. System Architecture

● In this system we will implement modules to provide security to home. This system developed base on machine learning approach.
We will implement module based on YOLOv4 and using neural network to achieve accuracyof module.
● Basically this system divided into two parts one for to detect unauthorized person and second module for to detect fire detection. In
this security application we will implement Mail service to send mail to house owner Using smtp lib.

• TO detect unauthorized person we will crate dataset of family member base on open we will scan user face and identify
unauthorized person face.
• In second part of system we will implement YOLO approach to detect fire in house .so we will harrcasecaded application to create
XML file of training module. We take dataset of fire from kaggle site.

IV. PROPOSED OUTCOME

● Fire detection through camera and notifies the owner through mail or msg or beep tone.
● Unauthorized can easily detect the and notifies the access person or owner. (in progress)
● Precious thing been taken other than owner then it will get notified to owner. (in progress).

IJRAR1CNP017 International Journal of Research and Analytical Reviews (IJRAR) www.ijrar.org 107
 © 2022 IJRAR April 2022, Volume 9, Issue 2 www.ijrar.org (E-ISSN 2348-1269, P- ISSN 2349-5138)
Output-

V. CONCLUSION
The venture of brilliant CCTV camera can be utilized for profoundly got regions or security of houses. Savvy video observation
system significantly adds to circumstance mindfulness. Such frameworks change video reconnaissance from information
procurement apparatus to data and insight obtaining frameworks. Constant video examination furnishes brilliant observation
frameworks with the capacity to respond continuously.

VI. REFERENCES

1. A Progress Review of Intelligent CCTV Surveillance Systems, Paper for IDAACS05 Workshop Sofia.
2. Ashwin S, Sathiya Sethuram A, Varun A, Vasanth P AMRITA School of Engineering, AMRITA VISHWA VIDAPEETHAM A
J2ME-Based Wireless Automated Video Surveillance System Using Motion DetectionMethod,
3. Tasleem Mandrupkar, Manisha Kumari, Rupali Mane, Smart Video Se-curity Surveillance with Mobile Remote Control2013,
IJARCSSE — 352 Volume 3,ISSN: 2277 128X.
4. https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=887 9591
5. (Basic Book/Monograph Online Sources) J. K. Author. (year, month, day). Title (edition) [Type ofmedium]. Volume(issue)
Available: http://www.(URL)
6. J. Jones. (1991, May 10). Networks (2nd ed.) [Online]. Available: http://www.atm.com

7. P. V. K. Borges and E. Izquierdo, “A Probabilistic Approach for Vision-Based Fire Detection in Videos,” IEEE Transactions on
Circuits and Systems for Video Technology, vol. 20, no. 5, pp. 721– 731, 2010.
8. Turgay Celik , Hasan Demirel, Huseyin Ozkaramanli, and Mustafa Uyguroglu, “Fire detection using statistical color model in
video sequences,” Proceedings of the International Conference on Acoustics, Speech, and Signal Processing, Jun 2006.
9. P. Gomes, P. Santana, and J. Barata, “A Vision-Based Approach to Fire Detection,” International Journal of Advanced Robotic
Systems, vol. 11, no. 9, p. 149, 2014.
10. T. Çelik, H. Özkaramanlı, and H. Demirel, “Fire and Smoke Detection without Sensors: Image Processing Based Approach,” 15th
European Signal Processing Conference, Sep. 2007.
11. C.-B. Liu and N. Ahuja, “Vision based fire detection,” Proceedings of the 17th International Conference on Pattern Recognition,
2004. ICPR 2004., 2004.

IJRAR1CNP017 International Journal of Research and Analytical Reviews (IJRAR) www.ijrar.org 108