Front of the Friendly Face Recognizer
Recognizer in use detecting a face
Back of the Recognizer
Have you ever felt too safe in our world of social media?
My original inspiration for this project was an episode of the Netflix series, Black Mirror, in which everybody has an ocular implant. One of the features of this implant is to view a live feed through somebody else's eyes, and anybody who comes into view is automatically searched on social media.
What it does
Our software performs facial recognition to quickly open a person's Instagram profile if the software sees their face.
How we built it
There are three main parts to the project: a camera device (we used a SnapDragon 410 Development Board), the recognition backend, and the phone link. The camera device utilizes OpenCV to watch for faces on the camera feed. Upon obtaining a stable, clear image of a face, it takes the image of the face and sends it to the recognition backend. The recognition backend is entirely hosted on Amazon Web Services. When the image is received, it is dropped into an S3 bucket before a Lambda function compares the image against an Amazon Rekognition Face Collection. If the face is found in the collection, it finds and sends the Instagram username to our IFTTT webhook. IFTTT will take this username, and send a clickable notification to the user's phone. The user can tap this notification to be taken to the person's Instagram account.
Challenges we ran into
We ran into many issues while compiling our code for the development board.
Accomplishments that we're proud of
We came up with a cool, yet creepy idea, and were able to plan it out and bring it to life in a relatively short amount of time.
What's next for Friendly Face
Spying on the masses... All joking aside, we had a few ideas that would require more time to complete, such as automatic Instagram scraping, facial recognition across profiles, and support for more social media sites.