Our team was inspired to find a solution for safer driving after considering the recent legalization of marijuana. Being impaired due to being high, drunk or tired is a very pertinent issue. Impaired driving not only puts the driver at risk but also risks the lives of everyone else on the road. If a driver was to be impaired and remain unresponsive behind the wheel, they would most likely cause an accident on the road.
What it does
Blink uses Microsoft's custom vision AI trained with photos of open and closed eyes. With a webcam, companies such as Uber, Lyft and Taxis can have service which monitor's their drivers (specifically whether they become impaired/unresponsive) if they drive a smart car. When Blink detects that the user's eyes have been closed for 5s, the system issues an alert using Twilio. Immediately the latitude and longitude are forwarded to an emergency contact of the driver, telling them the ID of the car in distress and the latitude and longitude of the car.
How we built it
Challenges we ran into
We ran into problems trying to parse the JSON files from the servers, we also ran into problems trying to set up automated HTTP get and push requests to the Jstore server.
Accomplishments that we're proud of
We're proud of what we were able to produce with the time and resources that we had, for most of us this was our first hackathon. We're proud of being able to successfully train the neural network to recognize faces with closed or open eyes and being able to communicate this information with the front-end and back-end seemlessly.
What we learned
We learned how to integrate various APIs and try many different ways to achieve the task and choose the best and the most efficient way to do so.
What's next for Blink
We would like to expand to notifying the nearest police department in relation to the GPS of the smart car. The police can then be notified that there is a potentially distracted driver in their area and get updates as to the location of the car. This would be done using Google's firebase and a few lines of code determining how to find the nearest police station to the current GPS.