Driver Drowsiness Detection System

Real-time driver drowsiness detection using OpenCV, MediaPipe Face Mesh, and a browser-based MediaPipe demo.

Live Demo

Try the browser version here:

https://hargunsiingh.github.io/driver-drowsiness-detection/

The live demo runs fully in the browser. Camera frames stay on the user's device and are processed client-side with MediaPipe face landmarks.

Features

Feature	How it works
Eye Closure Detection	Calculates Eye Aspect Ratio (EAR) from 6 landmarks per eye
Yawn Detection	Calculates Mouth Aspect Ratio (MAR) from lip landmarks
Head Tilt Detection	Tracks face orientation changes across consecutive frames
Desktop Audio Alarm	Two-tone pulsing alert via pygame in the Python app
Visual Alerts	Green/yellow/red state indicators, overlay landmarks, and warning display
Face Mesh Overlay	Live wireframe showing face oval, eye, and mouth outlines

Browser Demo

The deployable demo is a static site:

index.html
styles.css
app.js
face_landmarker.task

It can be hosted on GitHub Pages without a backend because webcam access and model inference happen in the visitor's browser.

Python Requirements

• Python 3.8+
• Webcam

Python Setup

pip install -r requirements.txt

Python Usage

python main.py

• Green border: driver is awake
• Yellow border: warning, such as yawning or brief eye closure
• Red border + alarm: drowsy, such as prolonged eye closure or multiple fatigue indicators
• Press q to quit

Configuration

Thresholds can be tuned in detector.py:

Parameter	Default	Description
EAR_THRESHOLD	0.22	EAR below this means eyes closed
MAR_THRESHOLD	0.75	MAR above this means yawning
HEAD_TILT_THRESHOLD	30 deg	Pitch above this means head drooping in the Python app
CONSEC_FRAMES_DROWSY	20	Frames of eye closure before drowsy alert

Project Structure

driver-drowsiness-detection/
|-- index.html            # Browser demo entry point
|-- styles.css            # Browser demo styling
|-- app.js                # Browser demo detection logic
|-- main.py               # Python entry point: webcam loop and UI
|-- detector.py           # Python detection engine
|-- alert.py              # Python audio alarm manager
|-- face_landmarker.task  # MediaPipe model
|-- requirements.txt      # Python dependencies
`-- README.md