Bad Mask

Inspiration

The theme of our hackathon was "quarantine", so we created a program that displays Youtube videos, or by default, Bad Apple, onto anyone wearing a mask in front of the camera.

What it does

It uses a facial detection system from the openCV library to detect for faces and masks, while also utilizing eye detection in order to locate the location and colour of any mask with a single solid colour. Using this solid colour as a greenscreen, we are able to display a video in the mask.

How we built it

We use code that barely hold itself together using packages that we barely understand through python to create this program.

Challenges we ran into

It was difficult to utilize facial recognition as none of us had done it before. We found openCV, a python library, but found it difficult to familiarize ourselves with its syntax and process. Through lots of learning on the internet, we were able to pull together some code that mostly works.

Accomplishments that we're proud of

The video displays. We were able to locate many features on the face, and our own database that was generated from a tiny sample size was sufficient enough to accurate tell if we were wearing a mask or not to a passable level of accuracy.

What we learned

Through this project, we learned a lot about openCV, computer vision in general, and machine learning.

What's next for Bad Mask

Over the course of making this project, we came up with several ideas that we would unfortunately not have the time to implement before the project deadline. In the future, we hope to implement these ideas and make Bad Mask an even better/worse product.