Scanning the Streets with a Manderson Grad

  • September 22, 2017

Stephen Buko

Stephen Buko of Kerb explains how the company creates representations of the streetscape.

Imagine that you’re a developer, looking to capitalize on the interest in mixed-use urban construction that integrates both the retail and residential. For these types of developments to be successful, there needs to be a critical mass of pedestrian foot traffic that will support the commercial businesses on the ground level. Plus, if a mixed-use development is fully-leased in the retail side with restaurants, boutiques, and bars, the apartments or condos upstairs are that much more attractive and can command higher prices.

There’s a bit of a problem, though, when determining where to build: How many pedestrians are there? Is a particular neighborhood full of young professionals walking about? Are there families? Pet owners? Deciphering the mix of the above can provide insights into whether a particular location is well-suited for development.

Kerb Technologies Logo

Kerb, co-founded by CEO Stephen Buko, a 2014 graduate of the Culverhouse College of Commerce’s Manderson Graduate School MBA program, is harnessing off-the-shelf tech and high-performance data analysis tools to provide a better sense of the streetscape.

Kerb’s approach to data collection is novel. Drivers for rideshare companies such as Uber and Lyft place cameras on the dashboards of their cars. The GPS-enabled cameras collect images as the drivers go down the road. The pool of visual and location data is then uploaded to Kerb’s cloud servers where it is crunched and a report is generated about the number of pedestrians in any given area. The algorithms are also sensitive enough to decipher if there are dogs or strollers, which according to Buko, “would help developers gain better insights into the block-by-block demographics and makeup of the neighborhood to a more granular degree than census data.”

The genesis for Kerb is Buko’s own experience living in the Dupont Circle neighborhood in Washington D.C. The notoriously bad parking situation in the area led him to look into what it would take to create a parking app. “Most parking apps take a crowdsourced data approach, which doesn’t work very well. I started looking into using computer vision to find parking, which is a much different approach,” he said.

What is computer vision? By assigning metadata to a pool of images, computers can learn to recognize objects or people through algorithmic data processing. This computer vision technique is being used to help driverless cars navigate the city. Buko explains, “If a driverless car sees what looks like another car in front of it, it swerves around it. If it sees what looks like a stop sign, it comes to a stop.”  

Kerb’s use of computer vision looks for streetside objects such as parking spots (or lack of parking spots), signage, and of course, people. Algorithms then parse and sort that data and spit out reports that provide exact details to the street-level fabric of an area.  

Buko’s background managing technical programs is aided not only by his MBA education, but also two engineering degrees from the University of Alabama: a MS in Mechanical Engineering and BS in Aerospace Engineering.

Currently, Kerb is selling its data to business improvement districts or BIDs. Buko says, “We’re looking to find trends in what vehicular or pedestrian traffic looks like on a Monday morning versus a Monday evening. What about Tuesday morning or evening? That’s what we’ll be able to offer”

This approach toward creating real-time urban analytics may eventually lead to web applications that tell exactly how many objects are in a particular area at that exact moment, which has major implications for a huge number of industries. One of those industries is advertising: Kerb’s technology can look for logos and brands, which for marketers, could be beneficial when deploying a major ad campaign in a new market.

However, he has extensive experience crossing disciplines and managing projects with different deliverables, which translates well to Kerb where, “Each of our potential clients want something different depending on where they’re located, but we’re able to apply the technology to provide them real solutions.”