Great Streets Pedestrian and Bicycle Counting PT 2

Sponsor: Department of Transportation, Great Streets Program

University: California State University, Los Angeles 

Status: Completed

Start and end dates: October 2017 (end date)

Link: Github 

This is a follow up project to the Great Streets Pedestrian and Bicycle Counting project. With this project we were able to help John Sam at DOT do automated counts with a different set of data, as well as receive a Toyota Mobility Grant. With the extra funding we hope to expand the project to a larger scale.


The Toyota Mobility Foundation (TMF) provided the City with a grant in the amount of $89,000 to implement a Pedestrian and Bicyclist Recognition Project. The initial partnership began between Great Streets and the Information Technology Agency (ITA) Data Science Federation, who selected California State University Los Angeles (CSULA) Data Science Research Group to pursue the work.

Leveraging the existing ATSAC network of video cameras at traffic signals, ITA and LADOT staff created video recordings of pedestrian and bicyclist behavior at selected intersections in the Arts District. The video footage was provided to the CSULA team, who have developed an deep learning to learn the built environment depicted in a video feed, and identify and count pedestrians and bicyclists who move through the camera frame, even tracking individuals throughout the frame so they are not double counted. Initial results were very promising, with accurate pedestrian recognition over dozens of hours of video footage from five different cameras.