Doomscrolling

An nose-tracking application using computer vision to detect and track nose direction to detect Doomscrolling in real-time.

Features

• Webcam Integration: Works with your device's webcam for live tracking
• YOLOv8 Model Support: Includes YOLOv8 nano and small pre-trained models for object detection

Technology Stack

• MediaPipe: Face mesh and landmark detection
• OpenCV: Image processing and video capture
• YOLO v8: Object detection capabilities
• Python: Core programming language

Project Structure

Doomscrolling/
├── yolov8n.pt          # YOLOv8 Nano model weights
├── yolov8s.pt          # YOLOv8 Small model weights
├── punishment.mp3      # Audio file
├── punishment.jpg      # Image file
└── README.md           # This file

Installation

pip install opencv-python mediapipe numpy

Usage

python Doomscrolling.py

The application will:

1. Access your webcam
2. Detect facial landmarks and iris position
3. Display real-time nose tracking visualization
4. Mirror the video horizontally for natural interaction

How It Works

The application uses MediaPipe's Face Mesh solution with refined landmarks to:

• Identify nose center points (landmarks 468 and 473)
• Provide pixel-coordinate mapping for on-screen interaction

Future Enhancements

• Add eye movement gesture recognition
• Integrate YOLO models for additional object detection
• Add real-time performance metrics

License

MIT

Author

AmrEladawey