InspirationWe are inspired by the struggle of many of us to keep safe during the current public health crisis.
That is why we have designed a system to keep track of Mask usage and to check that people use masks whenever they enter public spaces.
What it does
The system detects in realtime whether a person is wearing a mask and then responds with a mechanical output ie servo motor movement.
How we built it
Our system uses a Raspberry-pi camera to capture live images then sends the captured image to a custom HTTP Linux server. On the server, we use a TensorFlow model (from AIZOO) to examine the image from the Raspberry-pi. If the image contains a person with a mask the objection detection model will return true and send a request for the Raspberry pi to move a servo (“ie open the door”).
Challenges we ran into
We faced many challenges during the completion of the project. We tried to do all the image processing on the raspberry pi however the pi, unfortunately, did not have either the necessary speed or space to handle the complicated image processing in OpenCV and Tensorflow. Therefore, we had to design an HTTP server using python to send the image to a Linux computer where the image is processed. Finding the right model for mask detection was also difficult. We tried several different frameworks before finding an efficient Tensorflow mask detection model from AIZOO. Remote work also proved difficult especially for a hardware project where only one of us could see the immediate results.
Accomplishments that we're proud of
We are proud of integrating the hardware and ml software. There were many difficult aspects of the processing and sending the data between server and Raspberry-pi.
What we learned
We learned about the capabilities of different machine learning frameworks, the Raspberry-pi, and how to create an HTTP server and send images over this server.
What's next for MaskIt
We would like to extend our technology to test it to new situations ie large groups and with greater mechanical output ie doors opening. Hopefully, our technology could be beneficial in public spaces to check that visitors are using masks