International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 10 Issue: 04 | Apr 2023
p-ISSN: 2395-0072
www.irjet.net
ATM Security System Based on the Video Surveillance Using Neural Networks Prof. Manjunath Raikar1, Ms. Meghana S2, Mr. Prajesh3, Mr. Srichandan4, Mr. Zuhair Ahmed5 1Professor, Dept. of CSE, YIT Moodbidri, Mangalore, Karnataka, India 2B.E Student, Dept. of CSE, YIT Moodbidri, Mangalore, Karnataka, India 3B.E Student, Dept. of CSE, YIT Moodbidri, Mangalore, Karnataka, India 4B.E Student, Dept. of CSE, YIT Moodbidri, Mangalore, Karnataka, India 5B.E Student, Dept. of CSE, YIT Moodbidri, Mangalore, Karnataka, India ---------------------------------------------------------------------***--------------------------------------------------------------------surveillance detection. The research seeks to check for Abstract - Since many years ago, an unexpected or
several unusual activities, such as many people entering the ATM, removing cameras from the system, even if some camera masking is done, and even detecting people entering the system while wearing a helmet. When any of these circumstances arise at the ATM, the system immediately sends a notification to the closest station and locks the ATM door.
uncommon event detection has been utilized to identify strange elements in the collected data. The most popular method is machine learning, which plays a significant role in this field. In this essay, we've performed a review. on research that examines the machine learning model that uses these techniques to identify uncommon events. Our review is divided into four sections: usage of unusual detection, machine learning techniques, overall model performance, and classification of odd event detection. About 170 publications from research that was published between the years 2000 and 2021 and provides information on machine learning approaches have been recognized. After analyzing some of the research articles, we present 08 data sets which are included our experiment on odd detection as well as many additional datasets. By incorporating machine learning's unsupervised odd detection method, it is made more complex. Through this study paper, we encourage more studies based on the recommendations made by this review. There are numerous other experiments that use machine learning to identify odd events.
1.1 Objective Of Research The major goal of this system is to be proactive in ensuring public safety and preventing physical assaults by for seeing events. Traditional video surveillance systems rely too heavily on human oversight, which is exhausting and prone to mistakes. The sectors that require constant monitoring are expanding, and the danger of not spotting anomalies in time could lead to serious disasters. We were also inspired to take on this project to show how this extremely important activity can be automated and increase the efficiency of the surveillance system to alert any risks arising from the nondetection of anomalies in time as and when they occur. This id due to the advancement in the availability of sophisticated video cameras, technologies for continuous streaming of video data, and Deep learning techniques.
Key Words: Neural Network, Security, Object detection, Yolo, Haar Cascade
1.INTRODUCTION
2. LITERATURE REVIEW
For a long time, finding odd happenings was a big concern
[1]Using Neural Networks, anomaly detection in videos for video surveillance applications:
and problem. For various kinds of operations, there are numerous other ways being developed to identify odd events. The challenge of identifying patterns in data that weren't anticipated is known as unusual detection. The importance of identifying uncommon occurrences in various operational states aids in the identification of significant, sensitive, imperative, and practicable information. There is a need for automatic security warning systems in the current ATM systems, It makes it possible for users to enter the ATM safely. Even though the government and financial authorities have taken numerous measures to ensure safety, it is still costing fees for the human security system that are extra. The approach suggested here is a low-cost, real-time automatic ATM security system that is based solely on video
© 2023, IRJET
|
Impact Factor value: 8.226
Video labelers, image processing, and activity detection are the three layers that make up the entire process of anomaly detection in video surveillance. In light of real-time circumstances, anomaly detection in videos for video surveillance application provides reliable findings. Advantages include a 98.5% [1] accuracy rate at which the abnormality was found in photos and videos. The main drawback of this project is that it focuses on anomaly detection in terms of people's security
|
ISO 9001:2008 Certified Journal
|
Page 782