IoT Traffic Control using ML

• 

  vehicle detection

• 

  model procedure

Inspiration

We wanted to provide a solution to a growing local problem that is traffic in our town.

What it does

Our solution detects traffic and using stationary and aerial cameras. It processes the data in real time automatically controls traffic lights and diverts traffic accordingly to ease congestion.

How we built it

Using our personal equipment, a pre-trained model, a virtual server solution for live streaming the visual data.

Challenges we ran into

Live streaming the video from the cameras, real-time processing the footage with minimal delays.

Accomplishments that we're proud of

The vehicle detection is very accurate and the video streaming lag is minimal for the purpose.

What we learned

How hard it is to feed and process data in real time and how to pre-process data for efficient analysis.

What's next for IoT Traffic Control using Machine Vision

Being implemented in real world with more interconnected autonomous nodes that are able to decrease the problem of traffic with no human input.

Built With

• coco
• darknet
• nginx
• opencv
• pyqt5
• python
• pytorch
• yolo