A close friend of ours recently had an intruder in his house, which inspired us to come with a DIY technical approach to dealing with unknown people from entering.
What it does
Stranger Danger is a contraption that applies Computer Vision algorithms to train a model to recognize a list of known, preapproved faces. If an unknown face enters the premises and is captured by our high definition GoPro, a signal is triggered to activate a circuit of motors to propel a deadly plastic ball at the intruder. The intruder is also given a strong warning through speech-to-text.
How we built it
We used OpenCV to detect faces, and then trained a small model on images of known people. The mechanical aspect of the blaster was built using Lego Mindstorms and 3D printed holders and base. We trigger our circuit using a remote call to a raspberry pi over SSH which flips a servo to activate the blaster.
Challenges we ran into
This was our first time dealing with hardware and remote Raspberry Pi access, and under the pressure of time, we had to come up with fun (totally maintainable solutions) like having a script call a script which opens an SSH connection to call another script on the Raspberry Pi.
Accomplishments that we're proud of
We got it to fully function and are impressed with the demonstration of the project!
What's next for Stranger-Danger
We could incorporate new features like adding voice recognition to add users to our database if they say a safe word. A more robust face recognition model could also be trained.