Most of us went to the Microsoft Azure Blue workshop. We grew fond of the many facial recognition capabilities, and powers of facial recognition API's.
What it does
The project is meant to represent a form of home security. When a person walks towards the camera, the program determines whether or not the person entering the house is part of the family. The program then messages the home owner accordingly.
How we built it
Using IBM Watson's intelligent image recognition AI, we were able to train Watson to find and recognize different people. We used the Twilio API to perform the custom, automatic messaging to the user.
Challenges we ran into
Our greatest challenge was configuring the Visual Recognizer to distinguish between the different family members. This was because we were not able to use Custom Model Facial Recognition, so we instead had to train a Visual Recognizer that was not specialized in recognizing faces.
Accomplishments that we're proud of
Connecting Watson's feedback into a clean, readable interface on text is something we're all proud of (Thanks Alex!).
What we learned
We learned plenty about training and implementing machine learning API's. We also learned lots about how to properly set up a project, when dealing with unrelated API's.
What's next for WATSecure
We want to implement emotion recognition, and require the user to change his emotions in real time. This is in order to deter an intruder using pictures of the owner of the house.