Driver Drowsiness Detection System

Real-time driver drowsiness detection using MediaPipe FaceLandmarker and Eye Aspect Ratio (EAR).

Tech Stack

• Python 3.10
• OpenCV
• MediaPipe (Tasks API)
• NumPy
• SciPy

Methodology

• Face detection using MediaPipe FaceLandmarker (478 landmarks)
• Eye landmark extraction
• Eye Aspect Ratio (EAR) calculation
• Temporal smoothing using consecutive frame threshold
• Real-time alert generation when drowsiness detected

How It Works

1. Webcam captures live video
2. FaceLandmarker detects 478 facial landmarks
3. Eye landmarks extracted
4. EAR computed
5. If EAR < threshold for consecutive frames → alert triggered

Run Locally

python app.py