San Diego Confronts Bad Data From Bike-Counting Cameras
Monday, July 24, 2017
Photo by Guillermo Sevilla
At the intersection of West Grape Street and Harbor Drive, on the edge of Waterfront Park, there are two cameras mounted next to the north- and south-facing stoplights. The cameras look unassuming, but they are busy gathering data that will help the city plan its future transportation network.
Well, sort of.
The cameras, made by the Santa Ana-based company Iteris, are equipped with patented technology that can count bicyclists and extend green lights to give them more time to ride through an intersection. The cameras were purchased by the city of San Diego in 2014, but nearly three years later, they are still failing to gather credible data on how many people are biking through the intersection.
Technological barriers to counting bicyclists is a problem facing cities and regional transportation agencies across the country. But the need for good data on biking is especially dire in San Diego, where the mayor and City Council have committed to tripling the share of bike commuters in certain parts of the city by 2020.
"We need to monitor where we are in relationship to our goals," said Brian Genovese, a city traffic engineer who works on the city's bike program. "Right now our program is designing improvements to our bikeways, and we need to know if we're attracting an increased ridership based on those improvements."
City engineers have been watching the footage recorded by the cameras, counting bicycles and cross-checking their numbers with the automated counts gathered by the camera's software. The numbers do not match up.
Genovese declined to say how bad the data are, but he said the technology that distinguishes between bikes and cars is "evolving" and that he and his colleagues are still working with representatives from Iteris to troubleshoot the problem. Emails obtained via a public records request describe some cameras focusing on the wrong areas, or with dirty lenses.
Attempts to reach Iteris for an interview were unsuccessful.
The nearly three-year lag in trying to get the cameras in working order was due in part to staffing issues. The cameras were not installed until 2016, according to the city, because staff had a backlog of other work to do. It took even longer for the city to establish a live video link so staff could check the cameras' accuracy.
Samantha Ollinger, executive director of the nonprofit advocacy group BikeSD, said she was pleased the city was investing in bike-counting technology. But, she said, if the city had been quicker to install and calibrate the cameras, it could have three years of good data on bicycling.
"To not have good data, or to have buggy data that they aren't able to make good decisions on — it seems sort of a lost opportunity," she said. "I feel like a lot of other cities have solved the problem of how to count bike traffic."
Indeed, cities like San Francisco and Vancouver have set up much more robust bike counting programs. Long Beach, Portland and Montreal all have installed so-called "eco-totems" — electronic signs that display bike and pedestrian counts as they go up in real time. Ollinger said this kind of transparency is valuable, and can help encourage more people to bike.
Genovese said the cameras were purchased as part of a pilot program. But given the problems with the data, he could not yet say whether the program would be expanded.
"We're exploring different technologies," he said. "I think we'll have to explore those opportunities as they come up."
San Diego purchased dozens of cameras with bike-counting software in 2014, aiming to use their data to inform its transportation planning. But after a slow start to installing and calibrating the cameras, they are still failing to get accurate data.
To view PDF documents, Download Acrobat Reader.