I am really bad at remembering names, but Facebook seems to have an uncanny ability to recognize my friend's faces. Wouldn't it be awesome if we could use that technology for the greater good? Well, wait no longer! By combining Google Glass with said technology, we can create a powerful facial recognition device. We carefully reverse engineered the Facebook API and created a system that allows you to upload photos via html POST requests to remind yourself of people's names.

What it does

It takes a picture with Google Glass and uploads that photo to a server that we setup via Linode. This server then uses saved project credentials to load the image through a Facebook account and then runs Facebook recognition on that account. As long as you are friends with that Facebook account it should be able too recognize your face. A timetile is then displayed back to the user.

How we built it

We built WhoDatFace using Node, Google Glass, Linode, GDK, the Android API and other features. We carefully analyzed Facebook's upload process to discover the hidden endpoint used for facial recognition and utilized that to run the facial recognition service.

Challenges we ran into

Once we had the Facebook facial recognition working, integrating it with Google Glass proved rather difficult as the API is no longer officially supported and virtually no support can be found through more common means such as StackOverflow. We had to debug all the issues we encountered on our own, with minimal resources.

Accomplishments that we're proud of

We were able to reverse engineering the Facebook Facial Recognition API and construct a native Google Glass app that took advantage of the Glass hardware (touchscreen, camera, etc).

What we learned

We learned to reverse engineer APIs and how to make card and immersive apps for Google Glass through Android Studio and the Glass Mirror API.

What's next for Who Dat Face?

Bring support to more devices and making the network connection slightly more stable between the app and the server.

Built With

Share this project: