International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 10 | Oct 2024
p-ISSN: 2395-0072
www.irjet.net
SUSPICIOUS ACTIVITY DETECTION IN EXAM USING DEEP LEARNING Amar Kalukhe1, Siddharthsingh Suryawanshi2, Manish Ambuse3, Mahesh Ravaji4, Prof. Chaitali. A. Deshpande5 1,2,3,4Student,Department Of Information Technology, Sinhgad College Of Engineering, Pune 411041
5 Asistant professor,Dept. Of Information Technology, Sinhgad College Of Engineering, Pune 411041
-------------------------------------------------------------------------***-----------------------------------------------------------------------Overcoming Limitations of Depth Sensors: While low-cost Abstract: Suspicious Activity refers to the identification
depth sensors have advanced human pose estimation, their drawbacks—such as being limited to indoor settings and providing low-resolution, noisy depth data—are evident. This study proposes a new method that employs neural networks to overcome these challenges, leading to improved accuracy and robustness in human pose estimation.
of specific body parts or joint positions of individuals from images or videos. This project aims to detect unusual human behavior in real-time using CCTV footage by leveraging neural networks. The recognition of suspicious activities has been a significant challenge in computer vision for over 15 years. This area of study is crucial due to its wide range of applications, such as video surveillance, animal tracking, understanding behavior, detecting sign language, enhancing human-computer interaction, and enabling markerless motion capture. While low-cost depth sensors have limitations, primarily suitable for indoor settings, their low resolution and noisy data hinder effective human pose estimation from depth images. To address these challenges, we propose utilizing neural networks. The recognition of suspicious behavior in surveillance videos is a dynamic research field within image processing and computer vision. Through visual monitoring, human activities can be observed in critical public spaces such as bus terminals, train stations, airports, banks, shopping centers, educational institutions, parking lots, and roadways to deter threats like terrorism, theft, accidents, vandalism, and other criminal activities. Since it is impractical to monitor these areas continuously, intelligent video surveillance systems are needed to assess human activities in real-time, distinguishing between normal and abnormal behaviors while generating alerts. Most current research focuses on still images rather than videos, and there is a noticeable gap in studies employing convolutional neural networks (CNNs) for detecting suspicious actions.
Real-World Applications: This research specifically targets the recognition of suspicious human activity in video surveillance, a field that is rapidly evolving within image processing and computer vision. Through effective visual monitoring, human behavior in key public areas, including transportation hubs, retail spaces, and educational institutions, can be scrutinized. This proactive approach acts as a deterrent against a variety of threats, from terrorism and theft to accidents and other unlawful activities. The Need for Intelligent Video Surveillance: Continuous manual monitoring of public spaces is a significant challenge. This research advocates for the implementation of intelligent video surveillance systems that utilize neural networks for real-time observation and classification of human activities as normal or suspicious. These systems are also designed to issue alerts, enabling swift responses to any detected anomalies.
2. PROBLEM STATEMENT Detecting Suspicious Examination
Keywords: Suspicious Activity, Neural Networks, Image Processing, Surveillance Video.
Impact Factor value: 8.315
Behaviour
In
The Suspicious Activity Detection System (SADS) employs Convolutional Neural Networks (CNNs) for video analysis, specifically designed to enhance security measures in examination settings. Traditional monitoring methods often fall short in effectively identifying irregular behavior, making it essential to integrate advanced AI and deep learning technologies. SADS aims to provide a proactive and intelligent solution that exceeds conventional surveillance techniques. By harnessing the capabilities of CNNs, which excel at recognizing patterns and anomalies in visual data, this system can automatically detect suspicious activities that may go unnoticed during exams.
In today’s digital era, the importance of public safety has surged, with intelligent video surveillance systems playing a crucial role. The ability to detect and predict suspicious human activities from real-time CCTV footage has become a primary focus of research in computer vision and artificial intelligence. This paper investigates the various applications of neural network technology in addressing the significant challenge of detecting suspicious behavior, particularly within the context of video surveillance.
|
And
3. MOTIVATION
1. INTRODUCTION
© 2024, IRJET
Activity
|
ISO 9001:2008 Certified Journal
|
Page 772