RoadGuard AI

• 

  Not saving one person but their family

Inspiration

Every year, thousands of road accidents in India happen because of driver drowsiness and distraction. Most safety systems are only in expensive cars. We wanted to build something any driver can use with just a smartphone.

What it does

RoadGuard AI uses the phone camera to monitor the driver in real time:

• Detects drowsiness (eye closure, yawning)
• Detects distraction (looking away from the road)
• Plays an instant voice alert in the driver's language
• Saves a trip safety report

How we built it

• Face and eye landmarks detected with MediaPipe
• Eye Aspect Ratio (EAR) used to detect drowsiness
• OpenCV for video processing
• Python backend, simple web dashboard for reports
• Runs on a normal laptop or phone, no special hardware

Challenges we faced

• Accuracy in low light and with spectacles
• Reducing false alerts
• Keeping it fast on low-end devices

What we learned

Computer vision, real-time processing, and designing AI that is useful for real people.

What's next

Tamil voice alerts, hazard detection, and an app for fleet owners.

Built With

• computer-vision
• css
• flask
• html
• javascript
• machine-learning-"try-it-out"-links:-github-repo-link-(code-upload-pannitu-link-podunga)-demo-video-link-(youtube
• mediapipe
• opencv
• python
• tensorflow