Inspiration
Facial recognition is nice, but often can be inaccurate for different faces. Would like to accurately be able to find the features of a face with rapid relocation.
What it does
Finds your face, decides who you are, and can overlay images on your face.
How I built it
A little bit of machine learning, some numpy, scipy, opencv, and a bunch of secret sauce.
Challenges I ran into
Training the machine learning algorythm. Finding nice samples of faces that covered a broad spectrum of face types was a challenge. (I used the organizers faces)!
Accomplishments that I'm proud of
Making myself look like batman. Keeping a history of the different locations the features have been found in order to stabilize the tracking software as changes should happen gradually, as opposed to drastic changes.
What I learned
How to integrate many different open source libraries to optimize
What's next for Domecast
Mars, maybe relocating animals. Deciding if an animal has previously been found. I like to be friends with animals.
Log in or sign up for Devpost to join the conversation.