Inspiration

Realization of Internet of Things (IoT) gave me insight to different ways to solve complex problems in Nigeria. One greatest challenge among them is the deluge traffic that arises in almost every location at peak periods, using Lagos State as a case study.

What it does

The project is based on automating traffic management system, rather than the pragmatic way traffic light operates. we are building a traffic management system that autonomously controls traffic with the influence of sensors. Lets put it this way, you have have lanes intersection, with more 30 cars in lane A, 2 cars in lane B, 20 cars in lane C and 0 car in lane D. Traditionally, the way our current traffic system works, it may decide to show green on lane D having 0 car. But with our proposed solution, the traffic light autonomously give access to lane that has largest volume of cars. So any motorist that violates the traffic system, The traffic machine will snap the offenders' car plate number. this number will immediately get classified in a few seconds with our car plate number classifier. Details of the car owner will be phished out with our model and send down to appropriate law enforcement agency for further action.

How we built it

Firstly, we gathered demo plate numbers, trained along with demo labels (phone numbers and Name of owners). This datasets are supposed to be provided by FRSC, but since we are subjected to limited time, we made used of what are within reach.

We developed an object detector hardware. this hardware includes, camera, optical sensor, arduino, Led light (Red, green and Orange). This is detector is designed to control traffic by measuring volumes of cars on different lanes and give access to the lane that has highest volume of cars. Access provision to lanes were measure by threshold set via Arduino.

We purportedly include plate number classifier, to evaluate cars that attempt to violates traffic rules.

Challenges we ran into

Getting dataset to train our the model was challenging, because of time constraint. Since the only possible way to get car owner's plate number was through Federal Road Safety Commission (FRSC). To save time, we source for random plate numbers on Google and labelled them with random name (name of car owner) and phone number.

Another challenge was access to IoT hardware we used to control traffics from sensors that trigger switching of light based on number of cars, cameras and arduino.

Accomplishments that we're proud of

We were able to build autonomous traffic system that controls cars and identify defaulters. Which after classifier identify the car owner through plate number, it triggers a push notification mail into law enforcement agency.

What we learned

How deep learning and IoT connects

What's next for Automated Traffic Management System (ATMS)

  1. We need access to FRSC database, this will model to get accurate prediction of all car owners nationwide
  2. Solar power system that can keep the traffic system running without depending on the nation power supply.
  3. Sophisticated IoT tools
  4. Integrate solution with ARCGIS (to visualize location of all traffic system on Google)

Built With

Share this project:

Updates