acm-header
Sign In

Communications of the ACM

ACM TechNews

How Building Baseline Video Analytics via Crowdsourcing Can Lead to Safer Streets


View as: Print Mobile App Share:
A screenshot from software used by the Video Analytics Towards Vision Zero program.

he Video Analytics Towards Vision Zero project worksto eliminate collisions on roadways by training software to identify near-misses in intersections and understand the underlying road conditions.

Credit: City of Bellevue, WA

Bellevue, WA, Microsoft, the University of Washington, and the Institute of Transportation Engineers are collaborating on the Video Analytics Towards Vision Zero project, which is working to eliminate collisions on roadways by training software to identify near-misses in intersections and understand the underlying road conditions.

The program studies camera footage of vehicles, cyclists, pedestrians, and other road users in incremental clips to understand their behavior, gaining more skill at recognizing different roadway elements the more data it is fed.

The researchers have publicly opened the initiative so the algorithm can be taught to distinguish between different elements via crowdsourcing. In about a month, 500 people have volunteered to annotate existing traffic footage, and thus far the algorithm accurately recognizes motor vehicles more than 95% of the time.

Recognizing pedestrians is more difficult, and Bellevue transportation planner Franz Loewenherz wants multiple cities to access the tool to proactively correct problematic traffic corridors.

From Government Technology
View Full Article

 

Abstracts Copyright © 2017 Information Inc., Bethesda, Maryland, USA


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account