Authors - Vasu Agrawal, Nupur Chaudhari, Tanisha Bharadiya, Manisha Sagade Abstract - The recent progress in AI and deep learning has significantly transformed the public safety landscape, particularly in the area of real-time threat detection in public domains. This comes with increased complexity and density as urban environments become more complex; traditional surveillance systems are no longer enough for monitoring large crowds, detecting potential threats, or ensuring public safety. This has necessitated the development of automated systems that could process large volumes in real time to pick anomalies, suspicious behaviors, and objects liable to imperil security. We delve into the core methodologies that object detection models, such as YOLO, Faster R-CNN, SSD, and compare them .To further improve the accuracy in detection of anomalous and illegal activities and reduce false positives and negatives we created a custom dataset by fusing data from different sources, these systems enhance the overall reliability of the surveillance systems.
Open Zoom
