Computer Vision

Traffic Violations Identification

In addition to applications in drones, we are researching how computer vision can be utilized in transportation, particularly identifying traffic violations and helping instructors in driving schools.

Specifically, we use computer vision to process digital images and videos from traffic cameras, identifying lanes, road markings, traffic signs & lights, detecting violations and recording the offending vehicle's license plate.

To identify objects in the images and videos, we will use a variety of tools, which include Canny Edge Detection, Line Detection and the YOLO ("You Only Look Once") deep learning models.

These tools will help us identify the edges of the objects of interest (vehicles) as well as road markings and lanes, determine the direction of travel of the vehicles, and flag violations such as line crossing, driving on the wrong lane or in the wrong direction, as they occur in real-time.