Overview / Usage

Red light runners cause hundreds of deaths and tens of thousands of injuries each year. In 2016, 811 people were killed in crashes that involved red light running. Over half of those killed were pedestrians, bicyclists and people in other vehicles who were hit by the red light runners. In 2015, an estimated 137,000 people were injured in red light running crashes.

Methodology / Approach

To solve the problem described above I used the Intel Movidius and Tensorflow for vehicle detection and classification and speed prediction. The idea of this project is to detect cars that will not be able to stop in red light and update the traffic light signals to prevent cars from having an accident. It can double as an automatic speed ticketing machine.

The live feed of the camera is passed to Opencv to chop it into individual frames, then each frame is passed to the Tensorflow Model in the Intel Movidius to perform vehicle detection and classification and speed prediction. In the case of the camera detecting a car that is going to run a red light, the raspberry pi will communicate with the other traffic lights to not switch it to green.

Technologies Used

Intel Movidius stick, Raspberry Pi, Pi Camera, Tensorflow, OpenCV and Intel Powered PC.