Eric Piza, a professor of criminology and criminal justice at Northeastern University, says that camera tech has advanced so much in recent years that researchers can barely keep up.
“Camera systems are much more commonly incorporating AI and computer vision technologies, and cameras have become ingrained in police operations throughout the country,” he says.
Expanding Use Cases for Smart Cameras
With the help of CDW, Dawe designed an integrated solution leveraging Cogniac’s computer vision platform, Vantiq’s AI model and Meraki cameras to detect either humans or bubbles in the fountain. But training the model presented a challenge because the city couldn’t actually fill the fountain with bubbles without damaging it.
“We bought a little pool from Walmart and put it in the parking lot, and then we put a little green screen around it,” Dawe says. “We got a fire truck to fill it up with water, created all these bubbles, and then our marketing folks photoshopped those bubbles onto an image of the fountain.”
That gave Dawe what he needed to train the AI model, with a resulting accuracy rate of around 70%. Months later, when the fountain was scheduled to be drained and cleaned, Dawe got the green light to photograph bubbles in the fountain itself, upping the accuracy rate to nearly 85%.