Real-time Deep Learning Based Anti-Proxy Presence Monitoring

This study proposes a real-time presence monitoring system that leverages deep learning techniques to prevent proxy entries through facial spoofing. While facial recognition is widely adopted for its speed and non-intrusive nature, it remains vulnerable to fraudulent attempts using printed photos or recorded videos. To counteract this, our approach integrates the DeepFace framework for identity recognition with a real-time liveness detection module. This module evaluates dynamic facial features such as motion cues and texture variations to confirm the subject’s physical presence. If an attempted impersonation is identified, the system automatically flags it, ensuring only genuine individuals are logged. By combining accurate face recognition with effective spoof detection, the system delivers secure and dependable monitoring suitable for academic and professional settings. Experimental results indicate strong performance in both identity verification and live user detection across various real-time conditions.