Real time Drowsiness Detection

This project implements a computer vision and deep learning solution to detect driver drowsiness in real-time. Using CNN (MobileNet with Transfer Learning), the system classifies whether eyes are open or closed with high accuracy (AUC ≈ 0.99).

Key features:

Binary classification (open vs. closed eyes) from live webcam feed.

Real-time eye and face detection using OpenCV Haar Cascades.

Alarm system triggers if eyes remain closed beyond a set duration.

Achieved 98%+ accuracy, precision, and recall on test data.

This project demonstrates the use of TensorFlow, OpenCV, and deep learning for practical safety applications, especially in preventing accidents caused by driver fatigue.