Articles

Video Analytics Platform Helps Cities Monitor Pedestrians, Vehicles

Smart city startup Placemeter's platform can be paired with sensors for increased accuracy, and identifies the number and trajectory of five types of moving objects: pedestrians, bicycles, motorcycles, cars and heavy-duty vehicles.

by / June 1, 2016
The urban intelligence startup Placemeter has created algorithms that can identify five different types of moving objects after working with the city of Paris to redesign the Place de la Nation, a roundabout and plaza on city's eastern border. Google Maps

Digital image analysis isn’t anything new. Google and Facebook have used it to search and tag images, cybersecurity firms use it to detect criminals, and most recently, Twitter has applied it to its live streaming app Periscope to guide users to fresh content.

Despite its popularity, however, video analytics price points have often impeded cities from reaping the benefits afforded in the private sector. Until now, that is. Smart city startup Placemeter has invented an affordable video recognition technology that cities can leverage to analyze pedestrian and car traffic. The New York company’s co-founders Alex Winter, also its CEO, and Florent Peyre, its COO, said the platform is designed for nearly any video camera and installs in minutes.

“We actually spent a lot of time making this product extremely easy to use,” Winter said. “So in literally 5 minutes, you can connect your video feed and start measuring whatever you need to measure.”

The two said the advantages of their system is accuracy and accesibility: clear traffic information captured in real time that’s also easily shared. Transportation planners can refer to the data when designing transit routes, paramedics can optimize on-call ambulance locations, and police can schedule patrols based on foot traffic, to name a few.

"Its uses are pretty broad,” Peyre said. “But that's the exciting part, especially for cities, where just one stakeholder, or staff member, can be responsible for collecting the data, that can then be reused by many other agencies within a municipality."

The system, which can be paired with sensors for increased accuracy, identifies the number and trajectory of five types of moving objects: pedestrians, bicycles, motorcycles, cars and heavy-duty vehicles. The ingenuity in the classification stems from Placemeter’s algorithms, which enable what the startup calls “computer vision,” the analysis of image data patterns to identify objects.

Previously, automated traffic counters had to quantify moving objects based on where measurements were taken. Anything that crossed a sidewalk would be counted as a pedestrian; anything going through a bike lane was considered a bike, and any objects passing along a roadway were counted as vehicles. Placemeter’s founders say their system identifies moving objects no matter where they are, but does not collect personally identifiable information, such as facial recognition data or raw video footage.

Place de la Nation, ParisThe new features, released in May, were developed in concert with a major 2017 transit project in Paris. The city is redesigning a series of intersections, roundabouts and plazas for a more walkable, pedestrian-centered experience. Placemeter, which also is partnering with Cisco on the project, is focusing its urban intelligence service on the city’s Place de la Nation, a sizable roundabout and plaza on the city’s eastern edge.

“Object classification is key to this work [in Paris], so we can see how different designs for the space — including closing particular traffic lanes, moving benches and more — impact use by pedestrians, cyclists and drivers,” Peyre said.

In the U.S., New York City and Philadelphia have harnessed the technology for various analytics projects, and a partnership with Boston is expected later this year.

The grander vision for the technology is to construct a network of both real-time and historical mobility data for cities. The company hopes to partner with major urban landowners to who might assist in this kind of a distribution. In the short term, Winter and Peyre said they want to update the platform with algorithms that can identify additional objects, measure object speeds and calculate the length of time objects stay in particular places.

"The future, as we see it, is an endless flow of data that is quantifying human activity in cities in real time,” Peyre said. “It will be data that you have at the tips of your fingers."
 

Jason Shueh former staff writer

Jason Shueh is a former staff writer for Government Technology magazine.