Multi Authentication Security for Proactive Defense: Real-Time Suspicious Activity Detection Using Multimodal Fusion

Abstract

Public safety increasingly demands advanced surveillance systems capable of automatic and real-time threat detection. This study addresses this need by proposing a multimodal surveillance system that integrates face recognition, hand gesture detection, and body pose estimation to enhance threat detection accuracy. The system employs the MediaPipe algorithm to analyze video inputs in real time and promptly issue alerts when threats are identified. These alerts are forwarded to a mobile application linked to the camera network, allowing security personnel to respond immediately. Experimental evaluation demonstrates the system’s capability to improve situational awareness and reduce response latency. This approach offers a practical solution for enhancing public safety through the application of computer vision and real -time notifications.