Traffic systems in place around the city have long endured the obsolete method of "fixed" traffic lights and "fixed" traffic flow regardless of whether or not the street is occupied by a number of cars. We wanted to design a system that would streamline that flow and allow intersections and other busy streets to

What it does

By sensing its environment and the streets individually, in real time, it would generate information based on the number of cars present in the photo/on the street. It passes this information onward to the priorities check which would decide how to move traffic efficiently. While it is passing this information, it is keeping a track on the pedestrians that may wish to cross the street as well as the number of cars in each road/direction as if they reach a certain number of cars, the priority of the lights will change after the initial fixed allocated time.

How we built it

We trained the image recognition model using IBM Watson's REST API that would give us calls back and generate whether or not a car is present at that specific frame. We are generating our traffic light through the use of ARDUINO in order to switch the lights according to the rules and priorities we have set out. If one street's cars are more than the other, we would allow the heavier street to "STAY" green for a couple more seconds as it is adapting to the changing need of traffic congestion and vice versa.

Challenges we ran into'

Introducing Multithreading so that each thread may run a specific function as having an infinite loop within an infinite loop is impossible. We also ran out of glue in our glue gun because we were building a physical phone stand to meet the height requirements for the camera to take pictures of the street.

Accomplishments that we're proud of

Meeting every challenge we came across and completing them to the best of our abilities.

What we learned

Training a model needs patience and accurate execution.

What's next for StreetSmart

Taking into account historical data and letting the authorities know where the hot-zones are for traffic congestion.

Built With

Share this project: