Two student researchers at the University of Manitoba have teamed up with a Kelowna company called Two Hat Security to develop new software aimed at keeping child sexual abuse images off the internet.
It’s a growing problem, but with billions of images online, sorting through what’s appropriate and what’s illegal is a huge challenge for law enforcement agencies and social media companies.
Software which can scan and detect images previously reported to the authorities already exists, but the new software would be able to scan for images which exploit children as they’re uploaded.
Two Hat Security’s head of product development, Brad Leitch said the goal is to give police a new investigative tool and stop the spread of child sexual abuse images through social networks.
The RCMP received 14,951 reports of child sexual abuse images in 2015. The number of cases doubled to 27,361 in 2016.
"If you're talking about a new influx or a new trend of child sexual abuse material hitting the market there needs to be some kind of way to detect that,” said Leitch. "The ability to detect child sexual abuse material is probably the most meaningful because you're talking about the darkest place on the internet."
The RCMP hopes the software, known as computer vision, will help officers detect the illegal images and prevent kids from becoming a victim.
“Really where we’re going is the unknown material, where we have the opportunity to rescue child” said Sgt. Arnold Guerin with the Canadian Police Centre for Missing and Exploited Children. “What we’re trying to do is have computers learn what these images look like.”
Two Hat Security has enlisted the expertise of University of Manitoba student researchers Mehrdad Hosseinzadeh and Binglin Li.
“It’s very important I think for parents,” said Li. “They want their children to be safe.”
The PhD computer science students are fine-turning the software, so it can more accurately learn the difference between appropriate and illegal images. It’s a project the researchers feel passionate about it.
"I think the net should be a safe place both for children and for adults but more for children because they are more vulnerable to attacks and they are more sensitive,” said Hosseinzadeh. "I'm very passionate about the project. We're going to build a safer net as well as a safer world."
The students work is still in the very early stages, but Guerin said the RCMP eventually hopes to use the software to find images to prevent kids from becoming victims of exploitation.
Two Hat Security said the software would also be useful for social networks to detect and prevent child sexual abuse material from ever ending up on the internet in the first place.
The student research is being paid for by a government-funded agency called Mitacs.
The Winnipeg-based Canadian Centre for Child Protection said it's working on using similar technology.
While the technology could help protect children, the Canadian Centre for Child Protection said it addresses just one facet of a very large problem.