Smart Alarm System for Crime Hotspot Detection using Deep Learning

The increasing number of surveillance cameras to watch over human activities necessitates the development of a system that can automatically identify violent and suspicious events. This is an essential piece of equipment for smart buildings. The average yearly death toll from acts of human aggression worldwide is 7.9 per $\mathbf{1 0, 0 0 0}$ individuals. The majority of these acts of violence against people happen suddenly or in remote places. One of the main obstacles to halting these activities is the information delay in this case. In this work, the real time violence detecting technique is employed to thrive on this issue. Certain deep learning algorithms are applied in computer vision to anticipate and identify actions and attributes in videos. When police arrive at dangerous locations in real time, they analyze CCTV footage and begin an investigation before moving further. In smart buildings, this technology is mandatory. Violence Detection in real time will prevent any mis happenings from taking place by raising alerts to the security guards. Our model also has YOLO object detection inculcated in it in order to recognize any weapons or tools that an attacker may be using against people and the security guards will act accordingly.