Inspiration
As the topic is COVID-19, a disastrous problem for the world, I come up with the idea of doing something relevant to COVID-19 such as mask. Hence, the face mask detector is created
What it does
Detect whether a person is wearing a face mask or not, also detect whether a person uses hand to cover his/her face to fake a mask (the problem which was quite annoying and it makes a lot of noises was addressed) The model can detect in images, videos and real-time video streams There is also web demo by using Flask
How we built it
Using MobileNetV2 as base network, I train a balanced dataset of about 4000 images to detect a person is wearing a mask or not. I also used mediapipe to detect hand of that person and if the hand coordinate is in the "face area", it will be considered "No Mask".
Challenges we ran into
The model performs not good when we have many people and small face in the frame
Accomplishments that we're proud of
- Detect a person is wearing a mask or not
- Real-time
- Add good features to support the machine
What we learned
Computer Vision
What's next for Face Mask Detector
Deal with the problems we mentioned
Log in or sign up for Devpost to join the conversation.