Eric Jaffe in Atlantic CityLab: “A city window overlooking the street has always been a score in its own right, what with so many apartments stuck opening onto back alleys and dumpsters and fire escapes. And now, a company wants to straight up monetize the view. New York startup Placemeter is paying city residents up to $50 a month for street views captured via old smartphones. The idea is to quantify sidewalk life in the service of making the city a more efficient place.
“Measuring data about how the city moves in real time, being able to make predictions on that, is definitely a good way to help cities work better,” says founder Alex Winter. “That’s the vision of Placemeter—to build a data platform where anyone at any time can know how busy the city is, and use that.”
Here’s how it works: City residents send Placemeter a little information about where they live and what they see from their window. In turn, Placemeter sends participants a kit (complete with window suction cup) to convert their unused smartphone into a street sensor, and agrees to pay cash so long as the device stays on and collects data. The more action outside—the more shops, pedestrians, traffic, and public space—the more the view is worth.
On the back end, Placemeter converts the smartphone images into statistical data using proprietary computer vision. The company first detects moving objects (the green splotches in the video below) and classifies them either as people or as 11 types of vehicles or other common urban elements, such as food carts. A second layer of analysis connects this movement with behavioral patterns based on the location—how many cars are speeding down a street, for instance, or how many people are going into a store….
Efforts to quantify city life with big data aren’t new, but where Placemeter’s clear advance is its ability to count pedestrians. Cities often track sidewalk traffic with little more than a hired hand and a manual clicker and spot locations. With its army of smartphone eyes, Placemeter promises a much wider net of real-time data dynamic enough to recognize not only that a person exists but also that person’s behavior, from walking speed to retail interest to general interaction with streets or public spaces…”