03-14-2022 21:01 PM
Given that COVID-19 (Omicron variation) situation is still not yet controlled, the confirmed cases keep rising in some countries and cities. It's important to make sure people wear a face mask correctly to prevent a potential COVID-19 spread out. Most indoor areas are still required (compulsory required) to wear the face mask before entering. But it would become meaningless if people are incorrectly wearing the face mask. I am delighted to announce this Power App (Canvas App) template to help corporations, enterprises, schools, and universities check are wearing a face mask correctly before entering a venue.
This app is an infused AI model trained with a tool called "Lobe". It used a camera to take an image (every 1 second, it can be further tuned down) and perform AI model scoring to identify/classify continuously (every 0.5 seconds, it can be further tuned down) if that person is correctly wearing a face mask or not. If it detects people's face mask is wearing incorrectly, it will give the block sign. In reverse, it will give a green sign.
To download this Power App (Canvas App) template, please download it from my code repository on GitHub (easonlai/correctly_face_masked_recognition). Please give me a Star for this GitHub code repository and give me a Like below if you like it.
With my previous template of "Leave Office/Campus Safe" for tracking, we should make a more safe environment for people using a digital solution with Power Platform (Power App).
For how to import this template into your environment, please refer to this official article.
For more details on how to use an image classification model by Lobe in Power Apps, please refer to this official article.
I hope you enjoy it. And god bless, COVID-19 will be gone soon.
Below is a screenshot of sample images (Correctly & Incorrectly wearing the face mask).
I downloaded the template and loaded into Power Apps it but am not clear on how I get the model you trained in Lobe? Can you explain a bit more if there is a way I can get the trained model ( CorrectlyFaceMasked) or do I need to get the pictures from the repository and train a new model
I totally follow the Power App you made and am exploring LOBE integration with Power Apps at present but am not following how I get the trained Lobe Model that you created...
Thanks for your questions. Maybe you can refer to my blog post on Medium for more details about how to train a computer vision model with LOBE.
I built my own model for drinking water and got it working
I am studying what you did - you approached it differently than the app I am working on... Some similar things but also some different things .. will play around
Key learning from yours was using the .Classification rather than .Prediction that I was trying to use
I need to find the documentation on the dot ( .) notation for Lobe - let me know if you know where I can find that
Thx for the great demo!