🚗 Driver Drowsiness Detection System

Real-time Driver Monitoring System using Python, OpenCV and MediaPipe FaceMesh

---

📑 Table of Contents

• Project Overview
• Features
• Technologies Used
• Project Workflow
• Detection Parameters
• Performance
• Limitations
• Project Structure
• Installation
• Usage
• How It Works
• Libraries
• Roadmap
• Future Improvements
• Contributing
• License

---

📌 Project Overview

Driver fatigue is one of the leading causes of road accidents worldwide. This project provides a real-time Driver Monitoring System (DMS) that continuously analyzes the driver's facial landmarks using MediaPipe FaceMesh.

The system estimates the driver's alertness using:

• 👁 Eye Aspect Ratio (EAR)
• 😮 Mouth Aspect Ratio (MAR)
• 😴 Percentage of Eye Closure (PERCLOS)
• 🧠 Head Pose Estimation

If signs of drowsiness persist beyond predefined thresholds, the system automatically detects the driver's fatigue state and can trigger an alert.

---

✨ Features

• ✅ Real-time webcam monitoring
• ✅ Face detection using MediaPipe FaceMesh
• ✅ 468 facial landmark detection
• ✅ Eye Aspect Ratio (EAR) calculation
• ✅ Mouth Aspect Ratio (MAR) calculation
• ✅ Percentage of Eye Closure (PERCLOS)
• ✅ Head Pose estimation
• ✅ Driver state classification
• ✅ Lightweight and fast
• ✅ Modular Python implementation
• ✅ Easy to customize thresholds

---

🛠 Technologies Used

• Python
• OpenCV
• MediaPipe FaceMesh
• NumPy

---

🧠 Project Workflow

            Webcam
               │
               ▼
       Face Detection
               │
               ▼
    MediaPipe FaceMesh
      (468 Landmarks)
               │
      ┌────────┼────────┐
      ▼        ▼        ▼
    EAR       MAR   Head Pose
      │        │        │
      └────────┼────────┘
               ▼
      Driver State Analysis
               ▼
      Drowsiness Detection
               ▼
          Driver Alert

---

📐 Detection Parameters

👁 Eye Aspect Ratio (EAR)

Measures eye openness.

Lower EAR values indicate prolonged eye closure and possible drowsiness.

---

😮 Mouth Aspect Ratio (MAR)

Measures mouth opening.

Higher MAR values indicate yawning.

---

😴 PERCLOS

Percentage of Eye Closure over time.

PERCLOS is one of the most reliable fatigue indicators used in modern Driver Monitoring Systems.

---

🧠 Head Pose Estimation

Tracks the driver's head orientation.

Detects excessive downward or sideways head movement indicating fatigue or distraction.

---

📊 Performance

• Real-time Processing
• Approximately 25–30 FPS
• Detects 468 facial landmarks
• Low latency
• Lightweight implementation

---

⚠️ Limitations

• Works with one driver at a time
• Requires sufficient lighting
• Performance decreases if the face is heavily occluded
• Requires a webcam

---

📁 Project Structure

Driving-Monitor-in-Python/
│
├── calibration/
├── detection/
├── demo/
│   └── demo.gif
├── images/
│   ├── banner.png
│   ├── face_detection.png
│   ├── landmarks.png
│   └── alert.png
├── main.py
├── state.py
├── utils.py
├── face_landmarker.task
├── requirements.txt
├── LICENSE
└── README.md

---

🚀 Installation

Clone the repository

git clone https://github.com/badivana/Driving-Monitor-in-Python.git

cd Driving-Monitor-in-Python

Install dependencies

pip install -r requirements.txt

---

▶️ Usage

Run the application

python main.py

The application will:

• Open your webcam
• Detect facial landmarks
• Calculate Eye Aspect Ratio (EAR)
• Calculate Mouth Aspect Ratio (MAR)
• Estimate Head Pose
• Monitor driver fatigue
• Display alerts when drowsiness is detected

Press Esc to exit.

---

⚙️ How It Works

The Driver Monitoring System consists of three major components.

Face Detection

Uses MediaPipe FaceMesh to detect 468 facial landmarks in real time.

---

Head Pose Estimation

Calculates the driver's head orientation using facial landmarks.

---

Driver State Classification

Combines:

• Eye Aspect Ratio (EAR)
• Mouth Aspect Ratio (MAR)
• PERCLOS
• Head Pose

to classify the driver as:

🟢 Alert

🔴 Drowsy

---

📚 Libraries

• OpenCV
• MediaPipe
• NumPy

---

🗺 Roadmap

• Face Detection
• Eye Aspect Ratio (EAR)
• Mouth Aspect Ratio (MAR)
• Head Pose Estimation
• Driver State Classification
• Drowsiness Detection
• Audio Alarm
• Mobile Notifications
• Raspberry Pi Deployment
• TensorFlow Lite Optimization
• Multi-person Detection
• Streamlit Dashboard

---

🚀 Future Improvements

• Audio warning system
• Mobile notification support
• TensorFlow Lite optimization
• Raspberry Pi deployment
• Infrared camera support
• Driver identification
• Performance benchmarking
• Cloud logging
• AI-based fatigue prediction

---

🤝 Contributing

Contributions are welcome.

If you find a bug or have an improvement, feel free to:

• Fork the repository
• Create a new branch
• Commit your changes
• Open a Pull Request

---

📄 License

This project is licensed under the MIT License.

---

⭐ If you found this project useful

Please consider giving the repository a ⭐ on GitHub