iMissing

Landing Page
Uploaded Picture
FBI Poster Link
No Match Found

Inspiration

Every day people go missing. We wanted to build a service that could help find those people, rather than relying on more traditional methods of using amber alerts and what not.

What it does

iMissing is a website built on Flask that allows users to upload an image that will then be checked for any potential matches in the FBI Missing Persons database. If the image results in a match, the user is then redirected to the corresponding Missing Persons file in the FBI Database. Our technology can be used by law enforcement and regular people to identify suspected people. iMissing can also be extended in the future to pull data from other FBI databases, like the Terrorism, Most Wanted, and Capitol Riots database.

How we built it

We used HTML/CSS for the frontend, using Skeleton.css and Normalize.css for the boilerplate. We used Flask for the backend, and used the FBI API to pull data on missing people. We used the face-recognition library to recognize matching faces. A user basically uploads an image to a HTML form which posts it to the Flask backend. There, the FBI API pulls images of missing people from the Missing Persons database, and the face-recognition library is used to recognize matching faces. If there is a match, the user is redirected to a pdf that contains information on the corresponding missing person. If not, the user is informed that the missing person is not found in the database.

Challenges we ran into

We had little experience in Flask when we started the project, and there was a lack of documentation in both the face-recognition library as well as the FBI API. There were also a few false positives matches with the face-recognition library.

Accomplishments that we're proud of

We're proud of the fact that we could identify missing people with a reasonably high accuracy. Even as a prototype, this could be used to reunite missing people with their family and loved ones.

What we learned

We learned how to use Flask, as well as integrating the different parts of this website together. This was probably the largest hackathon project we've built so far.

What's next for iMissing

We could decrease the false positive rate by adjusting the threshold at which the face-recognition library decides it is a match. We could do more research on the library, or even opt for another one to improve the success rate of the machine learning model. Improving the machine learning model could also help with the runtime of this website. Lastly, we also want to extend our application to work with video.