While any one can be targeted by a trafficker, research has shown that traffickers often target children with increased vulnerabilities and prey upon a child’s vulnerability and use psychological pressure and intimidation to control, and sexually exploit, the child for financial benefit. Studies have found out that one in seven missing children (meaning anyone missing for over 24 hours) are likely to be victims of sex trafficking. After 48 hours it is extremely likely a child has traded sex for food, shelter or drugs.

What it does

Our tool allows the user to identify bias in AI related to human trafficking. The models we have created show increased sensitivities to minorities such as Hispanics and those aged one to five.

How we built it

We used IBM Watson Studio, IBM's Visual Recognition API, and AI Fairness 360 to develop custom models with machine learning specific to the National Center for Missing and Exploited Children.

Challenges we ran into

The IBM Diversity in Faces Dataset that we wanted to use is not currently available to the public.

Accomplishments that we're proud of

We are proud of our work to help combat human trafficking and increase awareness about this issue. Our path to social justice for victims of sex trafficking is long. We look forward to the day that communities come together to fight trafficking and the sexualization of children. We believe this tool will be part of that change.

What we learned

We learned about bias and fairness in AI and how it affects everyone. The IBM AIF 360 toolkit is helpful for reducing biases in models, specifically in facial recognition.

What's next for AI Detector for Missing Children

We want to add age progression to this tool so we can identify people that have physically changed drastically from when they were last seen. We also want not only to work with law enforcement within the Greater Boston area, but also local businesses, hotels and sports arenas to apply our tool. By gaining voluntary access to video footage during high trafficked human times such as the Super Bowl we may be able to save trafficked persons.

Built With

Share this project: