Related news
Discover more stories about Mitacs — and the game-changing innovations driven by students and postdocs.
Two computer science students at the University of Manitoba are tasked with developing software to curb the proliferation of child sexual abuse materials online without ever setting eyes on it.
“It’s a hard project, but as we go on blindfolded, to some extent, we will find a way to tackle the problem,” Merhdad Hosseinzahed said Thursday.
Hosseinzahed and Binglin Li are a pair of 26-year-olds from Iran and China, respectively, working toward a PhD as part of Project Cease, which is lead by Two-Hat Security from Kelowna, B.C.
Using advancements in computer vision — in short, analyzing digital images to extract data — the five-year program aims to go beyond current software that searches the Internet for previously reported CSAM and scan for images as they are uploaded.
The ultimate goal is to stop them from ever being posted, and to help authorities to quickly identify and prioritize new cases.
“It’s critical we continue to develop new technologies to address this multi-faceted, growing problem,” Lloyd Richardson, the director of information technology at the Winnipeg-based Canadian Centre for Child Protection, said.
Project Arachnid, a digital web-crawler launched by the centre this year, detected more than five million instances of CSAM in just six weeks.
The students were partnered with Two Hat by Mitacs, a non-profit national research organization that manages research and training programs in Canada using government funding.
Because the images are illegal to view, the students will build increasingly sophisticated dataset models for the RCMP to continually test for accuracy against its database.
“In this case, not only are we not allowed to see it, we actually cannot access it at all,” Yang Wang, a computer science professor at the U of M, said. “We basically have to do this somehow remotely, by some kind of magic.”
The project began in early January, with early work focusing on detecting the age of a person from an image. Later models will search for nudity, and the mood of the subject.
Despite the challenge, the students are eager to accomplish their goal.
“It’s very meaningful work,” Li said. “It’s about the children, and children are very important to the country and the future, and we want our children to be safe.”
Brad Leitch, head of product development at Two Hat, is confident the program will succeed on some level.
“Any success is going to be a win for law enforcement,” he said.
As a father of five young children, he’s also excited about the potential preventative aspect of the project.
“We can apply technology in a meaningful way to start fighting back,” Leitch said.
Kevin King