Building software tools to find child victims

Posted by Shumeet Baluja, Research Scientist

Since it was founded in 1984, the National Center for Missing & Exploited Children (NCMEC) has reported more than 570,000 child exploitation leads to law enforcement agencies and assisted with more than 140,900 missing child cases, resulting in the recovery of more than 124,500 children.

The advent of the Internet has unfortunately provided child predators with a new avenue to exploit children. In August 2006, we joined NCMEC's Technology Coalition Against Child Pornography, teaming up with other tech industry companies to develop solutions that hinder predators' ability to use the Internet to exploit children or traffic in child pornography.

In an outshoot of that industry initiative, I discovered some other areas where I thought Google could help the staff at NCMEC. For instance, to date, NCMEC analysts have reviewed more than 13 million child pornography images and videos to assist law enforcement agencies working to identify and rescue children. This task has been time-consuming, and NCMEC analysts were simply getting overwhelmed by all of the data they had to sift through.

One of our core strengths here at Google is our ability to manage and organize immense amounts of information -- whether it's text, image, audio, or video -- and make it more useful and accessible for users. As a member of Google's research group, I realized that NCMEC had an immediate need for some of our research-stage technology. They needed help organizing and making sense of the enormous number of images and videos sent to them every week through their CyberTipline and from law enforcement officers nationwide.

So we went into overdrive. I recruited some fellow engineers to help me build tools that NCMEC might find useful. Throughout 2007, using our 20% time, we created innovative software tools to help NCMEC track down child predators through video and image search. With these tools, analysts will be able to more quickly and easily search NCMEC's large information systems to sort and identify files that contain images of child pornography. In addition, a new video tool we built streamlines analysts' review of video snippets.

The keys here were organization, scalability, and search. In particular, the tools we provided will aid in organizing and indexing NCMEC's information so that analysts can both deal with new images and videos more efficiently and also reference historical material more effectively. We hope the tools we've built for NCMEC will help its analysts make the important and often time-sensitive work of investigating child predators faster and more efficient.

For me, working with NCMEC provided an incredible opportunity. It allowed us to immediately deploy some of our latest research in image and video analysis in a real-world setting. On a personal note, I've been truly inspired by the entire NCMEC team's dedication and diligence in completing such a heart-wrenching mission.

You always hope that your work will eventually be used do some good in the world, and this was an amazing chance to make that hope real by creating tools that have the potential to aid investigations of child predators, find child victims and reduce the flow of child pornography on the Internet.