Repping Canada in the back!
Mohammed's father lost a co-worker when he died driving when he fell asleep at the wheel. Falling asleep while driving is responsible for 72,000 crashes and 800 fatalities each year.
What it does
Using OpenCV and dlib (Neural networks pre-trained for facial recognition), we have developed script that can be run on the lightest of hardware. This script detects when your eyes are closed for more than 4 seconds and tries to alert the driver.
How we built it
We observed an example online where somebody made a script that could detect blinking. We have modified it to include sleep recognition and methods of keeping the driver awake.
Challenges we ran into
All weekend we pretty much couldn't install dlib (one of the neural net libraries). Eventually after moving everything to a virtual environment we were able to install it and get it working.
Accomplishments that we're proud of
Getting dlib installed.
What we learned
Go to hackathons with all your libraries installed.
What's next for dri(c)ve
Port it to android so that anybody could use it if they so desired. (And make it free).
If we wanted to sell it as a product we could develop a dedicated circuit board and enclosure. Then we could sell it to big motor companies like ford and chevy or to truck driving agencies. (We have proven that it can run on raspberry pi, so developing a custom board that uses an ARM processor wouldn't be that hard... see the schematics on my github to see how something similar has been done with an AtMega!!).
With love from (Newfoundland) Canada