According to an TOI article published in 2020, Sleep-deprived drivers responsible for 40% of road accidents, given that India tops the global accident and fatalities charts.

My goal is to introduce Artificial intelligence to solve this problem and create a platform that can keep an eye on the drivers and alert them if they show any signs of drowsiness.

What it does

It detects drowsy drivers and sends sound alerts as follows head down state - danger sound alert potentially sleepy state - warning sound alert awake state - no alert driver is safe

How we built it

I built it using browser based models via Tensorflow.js It uses MobileNet for a serverless backend It uses sound alerts implemented using pure Javascript.

Challenges we ran into

Identifying what should considered dangerous and potentially sleepy state. Making it mobile responsive. Since the idea was so unique and there is no webapp available like this i couldn't find a data set to train So custom manual training had to be used for now

Accomplishments that we're proud of

works as expected focuses on accident prevention instead of after accident injury prevention like the measures in place

What we learned

use Tensorflow.js and Mobilenet and interaction with intuitive minds

What's next for AccidentPreventionSystem

Scale it to make it as a product for transport companies as they have that are very exhausted and every accident for them is a loss of life as well as delay in the supply chain making the incident even more tragic.

Built With

Share this project: