Hacking at Microsoft //oneweek to ID Exploited Children

Oftentimes when a child is in trouble, one of the only clues we have is his or her face.  We can have a picture of a missing child and be looking for them, or we can have a picture of an abused child that was distributed online and want to find them quickly.  One of the main hurdles is how we connect the dots between these images of exploited children with other photos on the open web that may help us identify them and remove them from harm.  

We know that facial recognition technology exists and is used in other scenarios, but is has yet to be applied to this situation at scale.  This week, we’re excited to be joining the Microsoft team in Redmond at the Microsoft Oneweek Hackathon to tackle this challenge.  

Over the past several months, we’ve been exploring how to leverage Microsoft’s Cognitive Services to identify missing and exploited children.  We’ve had some initial positive results – enough to keep us going, but not nearly enough to build a solution that can scale.  So, this week, we’re working alongside some of the best and brightest Microsoft engineers and researchers to figure out how to work with large batches of images, how to deal with age progression and more.

In May, in two days at the Facebook cross-industry hackathon, we saw a lot of progress on related issues.  We’re excited about what will come of the next few days with Microsoft. Stay tuned!