-
cover
-
Drowsy
-
normal
-
Facial detecting coordinates
Inspiration
An estimated 1 in 25 adult drivers (aged 18 or older) report having fallen asleep while driving in the previous 30 days.
What it does
Detects if a driver is drowsy, in which case the app rings an alarm before the driver falls asleep completely. This will warn the driver to either pullover, or pay more attention on the road
How we built it
We used the opencv library to be able to map the whole face in terms of coordinates. We extracted eye coordinates that gave 6 specific points on the plane. Using the Euclidian geometry, we found out if the eye lid is closing with putting limitations in relation to the timespan it occurs. Not relying just on the coordinates and to be more precise and efficient, we trained a model using Microsoft ML, which analyses if our given frame is a frame with closed or open eye. The training set was images of people with eyes open and closed while the classifier predicts if the driver has his eyes on the road or not.
Challenges we ran into
Deciding out of the 6 coordinates, which anticipates the most accurately if the driver is feeling drowsy.
Accomplishments that we're proud of
Detecting facial landmarks from live video stream.
What we learned
Learnt a lot about openCV library.
What's next for U Snooze U lose
detecting drowsiness provided driver is wearing opaque glasses or under conditions when eyes are not visible.
Log in or sign up for Devpost to join the conversation.