🚗 Vision-based Driver Alert System

AI-powered vision-based system that detects driver drowsiness and triggers real-time alerts using facial landmark analysis

📌 Features

• 👁️ Eye Aspect Ratio (EAR) based drowsiness detection
• 😮 Yawn detection using mouth aspect ratio
• 📊 PERCLOS (Percentage of Eye Closure) monitoring
• 🔊 Real-time audio alerts (Text-to-Speech)
• 📁 Session logging (CSV + JSON reports)
• ⚡ High sensitivity and fast response
• 📷 Live webcam-based monitoring

---

🧠 Tech Stack

• Python
• OpenCV
• Dlib
• NumPy
• VLC (Audio Alerts)
• gTTS (Text-to-Speech)

---

🏗️ Project Structure

├── main.py ├── main_dlib.py ├── final-integration.py ├── shape_predictor_68_face_landmarks.dat ├── haarcascade_frontalface_default.xml ├── haarcascade_eye.xml ├── requirements.txt

---

⚙️ How It Works

1. Captures real-time video from webcam
2. Detects face and extracts facial landmarks
3. Computes:
   • Eye Aspect Ratio (EAR)
   • Yawn Ratio
   • PERCLOS
4. If thresholds are exceeded:
   • Triggers alert sound
   • Logs event
   • Displays warning on screen

---

🚀 Installation

git clone https://github.com/your-username/driver-alert-system.git
cd driver-alert-system
pip install -r requirements.txt

▶️ Usage python final-integration.py

Controls: Q → Quit R → Generate report

📊 Output Real-time EAR & PERCLOS values Drowsiness alerts on screen Audio warnings CSV logs + JSON report

📈 Key Parameters EAR Threshold: 0.15 Eye Closure Frames: 10 Yawn Threshold: 0.6

📂 Logs & Reports CSV file for event logging JSON file for session summary Includes: Duration Total alerts Drowsy events Yawn events

⚠️ Limitations Requires proper lighting Works best for a single face Needs webcam access 💡 Future Improvements Mobile app integration Night vision support Deep learning model (CNN/LSTM) Head pose estimation Cloud-based analytics

🤝 Contributing Pull requests are welcome. For major changes, open an issue first.

📜 License MIT License