International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 04 | Apr 2024
p-ISSN: 2395-0072
www.irjet.net
Smart Surveillance Systems: Detecting Anomalies for Enhanced Security Yashaswini B1, Shravya A2, Vemula Pallavi3, Srishti Mishra4 1,2,3,4Students, Dept of Computer Science Engineering, MVJCE, Bangalore, Karnataka, India
---------------------------------------------------------------------***--------------------------------------------------------------------The YOLO algorithm has been proven to be more effective Abstract -In today's fast-paced world, security is of utmost
than other detection algorithms. Compared to the Single Shot Multibox Detector (SSD), YOLO achieves a 3.2% higher mean Average Precision (mAP), while also reducing model parameters by 4.8% and computational requirements by 6.1%. By utilizing the YOLO algorithm, our system aims to improve security measures in various settings, from public spaces to workplaces, and ultimately contribute to the creation of safer and more secure communities.
importance, and we believe that surveillance systems are a crucial component in monitoring and responding to potential threats. Our project, which focuses on anomaly detection, plays a vital role in identifying unusual events such as falls, car crashes, and fires. Falls can often result in accidents or medical emergencies, while car crashes pose risks to drivers, passengers, and pedestrians. On the other hand, fires can cause significant damage and pose a danger to everyone in the vicinity.
1.1 RELATED THEORIES
Our system, which utilizes the YOLO (You Only Look Once) algorithm, integrates video surveillance and machine learning to automatically generate alerts when these anomalies occur. YOLO is a superior algorithm, achieving a 3.2% higher mean Average Precision (MAP) while reducing model parameters by 4.8% and computational requirements by 6.1%, compared to SSD. Our multi-modal anomaly detection system utilizes YOLO to improve security in public spaces and workplaces, promoting safer communities..
Our neural network has the ability to bring together the different aspects of object detection and form them into a cohesive unit. It uses characteristics from the entire image to forecast each bounding box and make predictions for all bounding boxes across all categories in the image at the same time(Hawkins, D. M. ,1980). This allows the network to reason holistically about the entire image and all its objects.. The YOLO design is a state-of-the-art object detection model that can perform end-to-end training and real-time detection while maintaining high accuracy. The input image is divided into a grid of S x S cells, and each cell is responsible for detecting any object whose center falls within it. For each cell, the model predicts B bounding boxes and their corresponding confidence scores, which reflect how confident the model is that the box contains an object and how accurate the prediction is. The confidence score is calculated as the product of the probability of an object being present in the box (Pr(Object)) and the intersection over union (IOU) between the predicted box and the ground truth. If no object exists in a cell, the confidence score is set to zero. However, if an object is present, the confidence score is set to the IOU value..
Key Words: Anomaly Detection, Alerting Mechanism, YOLOv8(You Look Only Once), Video surveillance, Falls, Car Crashes and Fire
1. INTRODUCTION In today's fast-paced technological world, ensuring people's safety and security is of utmost importance. Surveillance systems are essential in monitoring environments and responding promptly to potential threats(Chandola v, Bhaskar A, 2009). Anomaly detection is a vital component of these systems, as it helps identify unusual events or disturbances and alerts the relevant authorities promptly. This research project aims to develop a robust model for detecting specific anomalies, such as falls, car crashes, and fires. Falls caused by accidents or medical emergencies, car crashes that pose risks to people, and fires that result in uncontrolled combustion all require swift and accurate detection mechanisms. Our system uses the YOLO (You Only Look Once) algorithm, which is a cutting-edge approach in multi-modal anomaly detection. This algorithm integrates video surveillance with machine learning techniques, allowing automatic alert generation when anomalies are detected.
© 2024, IRJET
|
Impact Factor value: 8.226
Fig.1 Our system models object detection as a regression problem. The image is divided into an S x S
|
ISO 9001:2008 Certified Journal
|
Page 1235