International Journal of Scientific Research and Engineering Development

As digital devices become ubiquitous in shared academic, professional, and examination settings, concerns about screen privacy and unauthorized visual access (shoulder surfing) are growing. Traditional physical privacy filters offer static protection but fail in dynamic multi-user environments. This paper introduces GazeShield, a real-time intelligent screen privacy system that uses facial recognition and geometric head pose estimation to detect unwanted viewing. Identity is verified via 128-dimensional facial embeddings, and a Perspective-n-Point model analyzes yaw and pitch angles to flag suspicious gaze behavior. The system operates in four configurable modes: Single User, Member, Team, and Examination, triggering automated screen blurring alongside real-time alerts when intrusions are detected. Built as a secure Electron-based desktop application, it uses a FastAPI backend with JWT authentication, WebSocket based alerts, and PostgreSQL for logging violation events with timestamps and evidence snapshots. Testing under controlled indoor conditions with a recognition threshold of τ = 0.45 showed stable multi-face detection, real-time operation above 18 FPS, and reliable recognition confidence tracking via a sliding window analytics module. GazeShield fills the gap in existing work by unifying facial authentication, gaze-based intrusion detection, and automated protection in a single desktop framework, offering a lightweight, deployable alternative to complex gaze tracking systems with adaptive privacy enforcement across all four operational modes.

Mr. Sanjaykumar Ranveer, Vidhi Agre, Kanchan Bhange, Shruti Ghume,"GazeShield: Real-Time Intelligent Screen Privacy Protection using Face Recognition and Gaze Monitoring" International Journal of Scientific Research and Engineering Development, V9(3): Page(1297-1303) May-June 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.