The problem Project "Rush Hour" solves

It is a traffic management project, incorporating the fields of Machine Learning, Web Development and App Development. At each traffic signal, for each side, it counts the number of cars, scooters, buses, etc., and then assigns a corresponding priority and time for that particular signal, instead of just assigning a fixed time to all irrespective of the number of vehicles present at that time. Thus, this will save time of the commuters. Also, our project contains a special feature for ambulances, fire-fighters and other emergency-response teams to mark in their location of their starting and destination points and according to it, algorithms of all the traffic signals along that way will be altered in a certain way, allowing the emergency vehicles to pass the signal without any stopping at the signal.

Advantages:

1. Very less human intervention at each signal.
2. Quick and free roads for all the emergency vehicles, thus, allowing better social amneties for the citizens.
3. Better surveillance of the breaking of traffic rules, like overspeeding, wrong sides, not following the traffic lights, etc., because the same computer vision to count the vehicles, can also be used to detect them.
4. Overall monitoring increases coordination between different signals and helps in the betterment of one of the core features of a Smart City - Traffic Management.

Challenges we ran into

1. Integration of all the three tracks.
2. Slow output of ML Project because we were using masked RCNN, so, we used Tiny YOLOv3 Architecture for fast and real-time outputs.