Advertisement
Nasir Memon, professor of computer science and engineering at New York University, is leading the development of new filters that can detect child pornography in a more efficient manner, helping law enforcement catch child predators faster. Photo: Courtesy of New York University

The staggering amount of data involved in searching for child pornography is a serious hurdle in hunting down pedophile rings. Much of the image searching is automated, saving time instead of manually scouring millions of items of suspicious material.

Most still pictures can be digitally flagged for prime clues—even down to fingerprints and names on prescription bottles in the frame. But the accuracy of video, which varies in quality and consistency even within the same clip, has proven a more difficult task.

A new set of filters improving the ability to comb through terabytes of data and hundreds of thousands of hours of video is now under development by computer scientists at New York University.

The machine-learning techniques focus on picking out nudity—and identifying the physical features of children, said Nasir Memon, the NYU professor of computer science and engineering leading the work.

“The people who produce this content are clever—but they make mistakes,” said Memon, in an interview with Forensic Magazine.

The AI learning detects skin tones in low light and in poor quality footage. The machine also maps those skin tones to determine if someone in the frame is nude.

The other AI filter extracts facial features, and uses spatial and contextual analysis to determine whether the face and body belong to an adult or a child, he explained.

Thus far, the algorithms accurately detect nudity and explicit images 83 percent of the time, and identify children roughly 96.5 percent of the time, he added.

The tool, which does not yet have a project name, will complement existing mechanisms in Griffeye Analyze, a digital investigation platform used internationally.

Memon has previously worked in still image analysis. But video is the next frontier, considering the prevalence of smartphones, and what has been called an “explosion” of child porn flagged by agencies like the National Center for Missing and Exploited Children, he said.

The NCMEC work is especially valuable in light of the recent proliferation of sexual exploitation in cyberspace, said Memon.

“They’re doing amazing work—they’re saving a thousand children per year,” he said.

The NYU work is being funded by a three-year National Institute of Justice grant totaling $465,000. A working version of the suite is expected to be completed by the end of this year, and years two and three will be spent refining the tool, Memon said.

Johann Hofmann, the director of Griffeye, said that the NYU work is going to improve investigations around the world—and just in the nick of time.

“The scale of the problem, and the horrific nature of the material, means law enforcement officers are in great need of technology that can help them quickly prioritize data and detect contraband materials,” said Hofmann, in a statement released by NYU. “Using these novel child exploitation classifiers, developed by (Memon) and his team, together with the Griffeye Analyze platform, guarantees increased efficiency and better results in processing cases and identifying victims.”

The NCMEC Cyber Tipline is the biggest clearing house and conduit between child exploitation leads and investigators across the U.S. When Forensic Magazine spoke to Lindsey Olson, executive director of NCMEC’s Exploited Children’s Division, in November, the number of tips and leads had already topped 7.5 million—a record-setting total.

Olson said a combination of factors have led to the alarming rise of child porn. What may have existed 10 years ago in cyberspace has multiplied due to technological breakthroughs—but it’s also out in the open now, due to better cooperation among the internet community, she explained. 

“I think it’s the internet, smaller and cheaper storage devices that store more—all of these contribute to the increase in the amount of content available,” said Olson.

Advertisement
Advertisement