PRIVSIGHT

AI-powered desktop privacy shield with intelligent face recognition and notification management

Intelligent Privacy Features

👥

Shoulder Surfing Detection

Real-time detection of multiple faces. Automatically hides sensitive notifications when unauthorized persons are present.

👁️

Smart Attention Detection

MediaPipe-powered gaze tracking detects where you're looking. Hides notifications when you're not at your screen.

🏷️

Notification Classification

Intelligent app sensitivity rules. Classify apps as sensitive, moderate, or public with customizable filtering.

⚡

Performance Optimization

Lightweight and efficient with CPU usage under 10%. Real-time performance monitoring and adaptive frame skipping.

📹

Live Camera Preview

Real-time video streaming with face detection overlays. Monitor what the system sees with 15-30 fps streaming.

🔒

100% Local Processing

All processing happens on your device. No data sent to cloud. Your privacy is completely protected.

Performance Targets Achieved

6-8%

CPU Usage (Target: <10%)

140-180MB

Memory Usage (Target: <200MB)

30-50ms

Latency (Target: <100ms)

20-24fps

Processing FPS (Target: 15-30)

Built With Modern Technology

Frontend

✓ Electron 27 for cross-platform desktop
✓ React 19 with modern hooks
✓ TypeScript for type safety
✓ Real-time WebSocket communication

Backend

✓ Python 3.9+ backend service
✓ OpenCV for computer vision
✓ MediaPipe for gaze tracking
✓ face_recognition for identification

Getting Started

1. Install Dependencies

cd electron && npm install
cd ../python && pip install -r requirements.txt

2. Start Python Backend

cd python && python -m server

3. Start Electron App (in new terminal)

cd electron && npm run dev:renderer

4. Register Your Face

cd python && python register_face.py "Your Name"

✓ Ready to go! The app will start detecting faces and managing your notifications intelligently.