An Intelligent Anomaly Detection Management Solution Suitable for General CCTV Systems
Sun Kim, Geon Pyo Gwon, Seong Geon Yun, Yun Ki Hong, Hyen Ki Kim, Ki Hyun Jung · Asia-pacific Journal of Convergent Research Interchange · 2024
While the existing general CCTV system only provides a simple video recording function, intelligent CCTV includes reinforced functions such as real-time detection, object analysis, and automatic warning.Recently, the importance of introducing a large-scale intelligent abnormal behavior detection system has emerged due to the limitation that the existing CCTV system needs to satisfy the modern security needs in Korea.This paper proposes a more efficient way to perform security control in terms of cost without additional installation or replacement by establishing and installing an intelligent abnormal behavior detection AI system on general CCTV.In the proposed system, an integrated control solution is designed and implemented using an AI model that combines YOLO (You Look Only Once) and SlowFast.The proposed YOLO and SlowFast models can effectively perform real-time object detection and time-series data analysis in the experimental results.In particular, the YOLO model demonstrates exceptional performance in real-time object detection, while the SlowFast network proves suitable for detailed analysis of time-series data.In addition, the proposed system can detect abnormal behaviors based on data analysis collected through general CCTV in the real environment and transmit related information to a web-based application that provides a rapid response.As a result, this study is expected to improve the existing CCTV system and fulfill the efficiency of security control by introducing the advantages of intelligent CCTV simultaneously.