Using AI to avoid touching your face

#azure #tutorial #ai #webdev

📺 028 - Em Lazer Walker 🎙️ Frank Boucher

In this episode, Em is showing one of her last projects. She trained a machine learning model using Azure Custom Vision to detect when someone, facing the camera, is touching or not their face. All this project is also available in GitHub.

Em will starts by explaining what a good machine learning model, and explain why the one use, in this case, is only "good enough for a demo". We will cover different options available when building your model and the scenarios when those options should be used. Once created, we will train our brand new model using many images of faces for both of the scenarios: touching and not touching.

The second part of this demo is composed of a JavaScript webapp. We will walk through the code that does all the operations to grab an image from the video feed and send it to the Custom Vision service.

Finally, she will share a VR project, inspired by the Microsoft Flight simulator (coming out that day), that use the Bing map service to let us fly in any sky around the globe.