IoT - Traffic Video Surveillance System for Weapon Detection

In the previous few years, multimedia traffic has significantly increased. The Internet of Things (IoT) introduces new traffic and applications. SDNs improve network administration. When used in conjunction with SDN, artificial intelligence (AI) can offer solutions to network issues based on approaches for classification and estimation. In this study, we investigate how to enable delay-sensitive applications in IoT contexts by integrating deep learning techniques with Software-Defined Network (SDN) architecture. As a case study, real-time video surveillance applications serve as the basis for the construction and testing of a large number of deep learning-based models for weapon detection. To aid in the early detection and prevention of various crimes, as well as to reduce the bandwidth requirements by offloading the communication network from massive traffic transmission using the fuzzy technique, The best deep learning model is utilized by a proposed artificial intelligence model at the edge that is located in order to extract the first discovered video frames that contain weapons in order to ensure prompt transmission of this information to law enforcement. In other words, the best deep learning model is used to extract the first discovered video frames containing weapons for prompt transmission to law enforcement. Results indicate that the network may minimize jitter by up to 80% on average and losses by about 4% by taking certain actions.