Driver Drowsiness Detection

A real-time driver drowsiness detection system built with Streamlit and TensorFlow that monitors eye movements through webcam feed to detect drowsiness and alert the driver.

🚗 Overview

This application uses computer vision and deep learning to monitor a driver's eyes in real-time and detect signs of drowsiness. When drowsiness is detected, the system triggers visual and audio alerts to help prevent accidents caused by falling asleep while driving.

✨ Features

• Real-time Eye Monitoring: Uses webcam to continuously monitor driver's eyes
• Deep Learning Detection: Employs a trained CNN model to classify eye states (open/closed)
• Instant Alerts: Provides immediate visual warnings and audio alarms when drowsiness is detected
• Web-based Interface: Built with Streamlit for easy access through any web browser
• Face & Eye Detection: Uses Haar cascades for accurate face and eye region detection

🔧 Technology Stack

• Frontend: Streamlit
• Computer Vision: OpenCV
• Deep Learning: TensorFlow/Keras
• Real-time Streaming: streamlit-webrtc
• Audio Processing: pygame
• Video Processing: av (PyAV)

🚀 Installation & Setup

1. Clone the repository

   git clone git@github.com:Dhanush-M555/driver-drowsiness-detection-streamlit-app.git
   cd driver-drowsiness-detection-streamlit-app

2. Install dependencies

   pip install -r requirements.txt

3. Run the application

   streamlit run Hello.py

4. Access the app Open your browser and navigate to http://localhost:8501

📱 How It Works

1. Camera Access: The app requests permission to access your webcam
2. Face Detection: Uses Haar cascade classifiers to detect faces in the video stream
3. Eye Region Extraction: Identifies and extracts left and right eye regions
4. Drowsiness Classification: The trained CNN model analyzes eye images to determine if eyes are open or closed
5. Alert System: When eyes remain closed for a threshold period (10 consecutive frames), the system:
   • Displays a "Drowsiness Detected" warning on screen
   • Shows a Streamlit warning message
   • Plays an audio alarm

🧠 Model Details

• Model File: drowiness_new2.h5 (pre-trained CNN model)
• Input Size: 145x145 pixel eye images
• Classes: Open/Closed eye classification
• Cascade Files:
  • haarcascade_frontalface_default.xml - Face detection
  • haarcascade_lefteye_2splits.xml - Left eye detection
  • haarcascade_righteye_2splits.xml - Right eye detection

🎯 Use Cases

• Personal Use: Monitor your own alertness during long drives
• Fleet Management: Integrate into commercial vehicle monitoring systems
• Research: Study driver behavior and fatigue patterns
• Safety Training: Demonstrate the importance of staying alert while driving

🌐 Live Demo

Driver Drowsiness Detection Website

⚠️ Important Notes

• Ensure good lighting conditions for optimal face and eye detection
• The system requires clear visibility of both eyes
• Audio alerts may not work in all browser configurations due to autoplay policies
• This is a demonstration system and should not be relied upon as the sole safety measure while driving

🤝 Contributing

Contributions are welcome! Feel free to submit issues, feature requests, or pull requests to improve the system.