Since the corona virus started, everyone (or almost) is worrying about the fact that they can be affected. Actually in many countries, people are not allowed to access public places (supermarkets, hospitals, shops, shopping, restaurants, etc.) without a mask. So in those places, there are people checking everyone to find if they are wearing mask or not, and it's a lot of work. That's why I thought about a solution that will solve this issue.

What it does

This project that will focus on automatically detecting whether a person accessing a public space is wearing a mask or not, using images coming from cameras.

How we built it

It first analyses images received from cameras to detect faces. Then, it analyses the face when it is detected, to find out whether the face has a mask or not. if the face has a mask, nothing happens. But if not, an alarm will go off to signal that a person accessing or inside (the pharmacy for example...) doesn't have a mask.

Challenges we ran into

The main challenge was to find the relevant datasets to use. I tested many, but at the end when testing, the accuracy of the algorithm wasn't good (System not efficient), so I had to search hard.

Accomplishments that we're proud of

What I'm the most proud is what I learned from python. I am a Data Science Engineering student, so this project has been (for me) another proof that python and it's libraries are such powerful tools, and it remembered me why I chose this this fascinating field.

What we learned

I learned how to use the tensorFlow library. As a data scientist to be, I have used a lot of libraries, but never tensorFlow . So this has been the opportunity for me to discover this amazing tool. Also it was the opportunity for me to really deal with a deep learning project (until now all the Artificial Intelligence projects I had worked on were about machine learning).

What's next for Mask On Face Detection

The next step is to capture and save the images of people (that the system detected not wearing mask) and save them in a database to maybe use later. Also, I would like to integrate this module in an enterprise system.

Built With

Share this project: