In the age of COVID-19, it is immensely important for all of us to wear face-masks. Keeping that in mind, we have designed a solution that uses cutting edge computer vision technology to detect people who are not wearing masks.

What it does

We followed a two faced approach to the problem:

  1. Non-mobile deployment: this uses Computer Vision to detect specific facial features from a video feed (either CCTV or webcam). The model determines weather the humans in the frame are wearing face masks or not within milliseconds for each frame input and requires minimal computational resources for each iteration. This model can easily be scaled up to be used in public spaces like airports, ATMs and office work-spaces, as long as we have RGB video input.

  2. Mobile deployment: We developed a light mobile application that accomplishes exactly what the previous model does. The upshot of this mobile based approach is that it is ultra portable, which gives it the potential to be deployed in a wide range of places like commercial cabs, temporary quarantine centers, hospital wards etc.

How We built it

  1. The first group was responsible for the Computer-Vision part:

    • We learned how to handle RGB video input in Python using open-CV
    • Searched for the relevant haar-cascade files online.
    • Did some trial and error with the cascade files.
    • Then we integrated the haar-cascades into Python code.
    • We checked a few corner cases, for example when there are no people/multiple people in the frame.
    • We also had to make sure the model performs with minimal computational requirements.
  2. The second group was responsible for the Mask Detector App:

    • We started by researching about the mask detection feature and then came across the path to follow.
    • First designed how Our App will Look and What all features it will contain.
    • Then divided the tasks among our teammates and worked towards achieving the required results.
    • At last all our efforts aligned and we were able to get an outcome.

Accomplishments that We are proud of :

  • This was the very first hackathon for all our teammates and we are extremely pleased that we started our journey with MLH NewFriends NewHacks.
  • We developed team-spirit among ourselves and also learned to divide the tasks and effectively complete each of them to achieve our required results.

What's next for Mask-Appeal:

  • Improving the current accuracy of the model that has been built.
  • Decrease the frame-rate to make sure it consumes low resources.
  • Create a mechanism to save mask-wearing logs in the storage.
Share this project: