Skip to content

Driver Drowsyness Detection experiement performed on MediaPipe and DeepLearning using Transfer Learning Technique

Notifications You must be signed in to change notification settings

codenigma1/Drowsyness-Detection

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

27 Commits
 
 
 
 

Repository files navigation

🚗 Drowsiness Detection System

A robust Drowsiness Detection System using two approaches:

  1. Deep Learning with Transfer Learning (InceptionV3) 🚀
  2. MediaPipe for Eye Aspect Ratio (EAR) and Head Pose Estimation 📈

This project aims to enhance road safety by detecting driver drowsiness in real-time using computer vision and deep learning techniques. The system alerts drivers when drowsiness is detected to prevent accidents.


🛠 Features

  • Real-Time Detection 📹

    • Uses webcam feed for real-time monitoring to identify drowsiness.
  • Deep Learning Approach 🧠

    • Robust and Flexible: The transfer learning model with InceptionV3 is trained to detect drowsiness under various conditions. It can accurately classify eye states (open/closed) even with head movement, moderate lighting changes, and different face orientations.
    • Suitable for applications where precision and adaptability are crucial.
    Deep Learning Approach Demo Deep Learning Approach Demo Deep Learning Approach Demo Deep Learning Approach Demo
  • MediaPipe Approach 🧩

    • Lightweight and Efficient: This method calculates the Eye Aspect Ratio (EAR) and monitors head pose without the need for neural network training.
    • Limitations: It works well when the neck is not excessively tilted down, making it ideal for scenarios with minimal head movement.
    • Highly efficient, suitable for resource-constrained devices.
    Deep Learning Approach Demo
  • Custom Alerts 🔊

    • Plays an alarm sound if drowsiness is detected for an extended duration.
  • Interactive Visualization 🖼

    • Displays live status, EAR values, and predictions with overlays in real-time.

📊 Results and Insights

Deep Learning Approach:

  • Achieved high accuracy on test data with fine-tuned InceptionV3.
  • Robust to variations in lighting, face orientation, and head movements.
  • Ideal for systems requiring precision across diverse environments.

MediaPipe Approach:

  • Lightweight and efficient for devices with limited resources.
  • Effective when head movements are minimal and neck tilt is not excessive.
  • Suitable for real-time, resource-constrained applications.

📬 Contact

For queries, feel free to contact me at: 📧 [email protected]

About

Driver Drowsyness Detection experiement performed on MediaPipe and DeepLearning using Transfer Learning Technique

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published