1 of 6

DriveSafe

AI-Based Real-Time Driver Drowsiness Detection and Emergency Alert System

Team O(1) Visionaries

​

Aryan Sharma — Team Leader & Researcher

Sumit Kumar Pandey — Backend Developer & Deployment

Riya Gupta — Integration & Testing

Ashika Sanraj — Frontend Developer

​

PROJECT IMAGE

DriveSafe is a camera-based driver monitoring prototype that detects signs of drowsiness through eye monitoring and provides a warning when prolonged eye closure is detected.

​

2 of 6

THE CHALLENGE

Drowsiness can change a driver’s reaction before they notice it

Fatigue can impair attention and driving performance.

deaths from drowsy-driving-related crashes in 2024

Source: NHTSA, 2024

Target users: private and commercial drivers, night-shift and long-distance drivers, fleet operators, and road-safety organizations.

DriveSafe explores an accessible software-based way to detect a warning sign early.

3 of 6

THE CORE LOOP

We built the detection-to-warning loop

01

Monitor

Live camera monitoring captures video frames.

02

Locate

MediaPipe facial landmarks locate the eye region.

03

Estimate

Eye Aspect Ratio estimates eye openness over time.

04

Warn

A configured closure threshold triggers a warning.

4 of 6

SYSTEM FLOW

Four stages turn camera input into an alert

Camera Input

Eye Landmark Detection

EAR Analysis

Drowsiness Warning

STACK Python + OpenCV · MediaPipe · EAR · React · Spring Boot architecture · planned MySQL persistence

5 of 6

LIVE DEMO

The live demo proves the core workflow

01

CAMERA VIEW

Editable placeholder

Insert the real DriveSafe camera screen.

02

EYE DETECTION

Editable placeholder

Insert the real face and eye landmark capture.

03

WARNING SCREEN

Editable placeholder

Insert the real drowsiness warning capture.

WORKING NOW

Live camera monitoring, eye detection, and drowsiness warning.

NOT YET ESTABLISHED

Formal accuracy, false-alarm rate, or accident-reduction impact.

6 of 6

TECHNICAL CHOICE

EAR gave us a simple, explainable technical choice

Landmark Geometry

Map key eye points to compute vertical and horizontal distances.

Aspect Ratio

Compute EAR = (vertical sum) / (2 * horizontal) to quantify openness.

Temporal Monitoring

Track EAR across frames to detect sustained closure events.

Configurable Threshold

Set threshold and consecutive frames for model-free detection.

Trade-off

Lighting, glasses, camera angle, and individual differences can affect detection.