Statistics Canada has launched the Traffic Flow Dashboard as experimental statistical information on the traffic flow on selected road segments in the cities of Calgary and Toronto.
This dashboard shows traffic count data that is obtained from traffic camera imagery using a computer vision-based system developed in the Data Exploration and Integration Lab (DEIL) at Statistics Canada. The system periodically extracts traffic flow images from the Application Programmable Interfaces (APIs) of municipal and provincial traffic camera programs. Vehicle detection was implemented using the open source You Only Look Once version 3 (YOLOv3) object detection model. The output is used to generate real-time counts of the detected vehicles (cars, trucks, buses, motorcycles).
More information on the methods used to generate traffic flow estimates can be found at the link below.
Traffic volume estimation from traffic camera imagery: Toward real-time traffic data streams