Design of Neural Network-Based Smart City Security Monitoring System

Design of Neural Network-Based Smart City Security Monitoring System

연구 분야: Infrastructure

학회: ICCMT '24: Proceedings of the 2024 International Conference on Computer and Multimedia Technology

This paper presents a smart city security system using AI to enhance urban monitoring and public safety. It features a hierarchical architecture, processes video and multimodal data, and employs CNNs and RNNs for detecting 173 types of violations with 93.6% accuracy. The system, hosted on Kubernetes, shows excellent scalability, with a query response time of 246.8 milliseconds and 98% of events processed under 1 second. Future work will focus on improving complex scene recognition and optimizing GPU resources.