Mask Detector to prevent covid-19

as in 10 April 2020 the Indonesia Government has regulated massive social and physical distancing rule and starting in Jakarta along with few other big cities to prevent the spreading of covid 19 virus. And the most important rule is that every people who are still working and still use public transportation or services are obligated to use Medical Mask or regular mask in any kind.

While physical distancing is the most important thing to prevent picking up the virus, wearing face mask could also limit the transmission via mouth and nose.

Also patient who have the virus are highly required to use the mask everywhere they are.

Medical worker who's handling the patient also are required to do so.

So based on those things, I am creating this project that could be implemented in hospital and other public places where people can be crowded

Step 1 - Prepare your hardware

since I am using sipeed maix bit which use the kendryte K210 processor that mean you can use other sipeed maix hardware that use the same processor

I am using the kit that already include an SPI LCD and camera. I Modify the box that come with it make an enclosure case for the hardware

And I also add a small breadboard to held the monitor and the microcontroller which became like this

That's it for the hardware, since this project is focus on the AI side there's not much for the hardware component

Step 2 - Getting to know the maix sipeed ecosystem

There are few ways to use the sipeed maix board, but for this project I am using the micropython version which has detailed documentation here
(other reason also because I am a micropython geek)

for video introduction you can check dmitry's from hardware ai video

So I am not gonna make further getting started steps because on the link above already have so much things regarding the hardware

Step 3 - Making a custom data model

there are a few ways to make data model for this hardware including converting a tensorflow, yolo, or even keras. like this video

but since I am not having a powerful computer to make an accurate data model, sipeed maix already have a cloud platform to make a custom data model called Maixhub

before we go to maixhub, we need to collect a mask dataset and a no mask data set which you can grab both of it from kaggle.

Mask Dataset

No Mask Dataset

After you download both dataset, now you can go to Maixhub and train your data. More details about how to make datamodel and launch it on your board is explain in the video below from caesar wu (the creator of sipeed maix)

Step 4 (Skip Step 3 if you want to)

If you don't want to make your own data model, you can use my data model that I already put in the link below. Just make sure you watch the video on Step 3 to have a better knowledge on how to use my data. This will save you a lot of time.

Step 5 - Result

So here is the result of data model That I create

You can add a lot of other feature such as, notification when no mask detected, Statistic of mask user in a certain area for assuring safeness in a certain area. Also there a new law in Indonesia for people who go outside without wearing a mask, this project could detect the person who broke that law.

Hopefully this project will help in this weird time and situation also I hope all of you are healthy and virus free

Credits

Tutorial video are provided from:

Hardware AI Youtube channel

Caesar Wu Youtube channel

Datasets are from:

Kaggle User Niharika Pandit and Mikolaj Witkowski

Cover photo taken from asiatimes.com