Virginia company uses AI to detect potential gun violence – NBC4 Washington


No matter where people go all day, they are likely to be captured by the camera – at work, at the bank, at school, or at the mall. And sometimes those cameras catch a crime, like a shooting. But what if these cameras could help before the trigger was pulled?

One Virginia security company said it could.

Dave Fraser, CEO of Omnilet based in Leesburg. His company uses artificial intelligence to monitor video cameras for potential violence.

He said, “He finds out about pistols and rifles by basically taking the human’s place. By watching video channels. Does he do it 24×7.”

There are an estimated 70 million security cameras in the United States, but according to one industry survey, less than 1% of them are monitored by a human.

Fraser said his producer is always on the alert.

“Don’t take a coffee break, don’t get tired or lose interest,” he said.

Fraser said the AI ​​technology uses security cameras that most companies already have, potentially keeping costs low, and is already being used in locations across the country and in the metropolitan area.

“In this area we have schools, we have malls, and we have some banks that use them,” he said.

Joe Belfiore is one such client. A retired DC police officer of 13 years now working in security, he’s seen his share of gun violence.

He said, “Everyone has security patrols, everyone has video cameras. How well do they work together?”

He’s had the technology running for a few months on some of the cameras in the apartment complexes he works in the area.

“I would describe it as, like, an early warning system for the possibility of armed violence, like a smoke alarm for gun violence. You have a picture of, you know, who might be carrying a gun somewhere. So, this is really helpful.”

Fraser explained how the technology works with News4.

“The red box indicates that the gun has already been discovered,” Fraser said.

Once the technology detects a gun, Fraser said, it passes that information within seconds to the designated operator who confirms the threat and activates the safety plan. This can include calling the police, sending warnings via text messages, emailing with signs or sirens, and even closing doors at the facility.

When asked how AI can tell the difference between a potential shooter and a police officer entering with a weapon, Fraser explained, “Technology isn’t just about recognizing weapons. It actually needs to identify that there is a threat.”

“Actually the gun should be brandished in order for it to be identified as a threat,” he said. “There are a lot of people who open carry or law enforcement officers, security guards.”

The technology is also trained to identify other common items that might look like a gun, he said.

“Ninety percent of the time a phone or something else is in people’s hands,” Fraser said.

I would describe it as, like, an early warning system for the possibility of armed violence, like a smoke alarm for gun violence.

Joe Belfiore, Security Specialist

During the Belfiore pilot, he said the technology has already gotten some false positives out of cell phones. He added that he has gotten smarter over time.

A company spokesperson told News4 that false positives are tested by all detection technologies and that Belfiore “was an early user of Omnilet Gun Detection, and there have been countless updates to the AI ​​during that time.”

Over the summer, he said, artificial intelligence discovered a weapon in one of Belfiore’s buildings.

He said, “I think it could focus attention and resources on fermentation problems. Yes, so I see it as potentially very promising.”

Fraser said that while no active shooters have yet been stopped using the new technology — hundreds of rifles have already been discovered.

“As it is used more, we have more intelligence to be able to train it and make it better and better,” he said.

Directed by Shawn Yancey, produced by Rick Yarborough, photographed by Steve Jones and edited by Jeff Piper.


Leave a Reply

Your email address will not be published. Required fields are marked *