Real-time Crime Detection Using Customized CNN
S. Samundeswari, M Harini, T Dharshini, S Srinithi · 2022 1st International Conference on Computational Science and Technology (ICCST) · 2022
Every day, there are more crimes committed and criminals are on the loose, which is making people fear for their safety. The primary goal is to detect and deter illicit activity before it occurs. With the aid of cutting-edge technology, CCTV is commonly used in both private and public spaces. It is possible to control crime in this area, but human supervision is required to oversee it. It's difficult for a human to keep track of multiple screens at the same time. Human error is a possibility in many situations. To overcome this drawback, we stipulate a Deep Learning-based Real-Time Crime Detection Technique that analyzes real-time CCTV footage and alerts a nearby supervisor about the crime in the current region. The model tracks the movement of people and classifies it as aggressive or nonviolent behavior using the Multiple Object Detection with Localization technique. Any aggressive conduct filmed by the camera will be detected and instantaneously alerted by the system.