Don't Touch Men (Mouth Eyes Nose)

This app warns you when your hand comes close to your face.

Comment

Inspiration

With Don't Touch Men, I am focussing on a particular issue related to the spread of COVID-19: Avoiding to touch your face. During video conferences and on social media I noticed a lot of people are touching their faces without realizing it. I developed a prototype to nudge people away from this behavior.

What it does

Its function is basically to alert you when your hand is near your face. It runs in the browser, Chrome recommended.

How I built it

It is based on a Tensorflow machine learning model (BodyPix). More information is on the repository on Github.

Challenges I ran into

I am currently switching from the first quick prototype to a faster and more reliable detection.

What's next for Don't Touch Men (Mouth Eyes Nose)

To go from prototype to a wider audience I would look for roles like Data Scientists and Product Managers (how can we make a product out of this), Designers (how to educate people), Mobile / Backend / API developers (to include the detection into videoconferencing products), Testers.

Built With

  • tensorflow.js
Share this project:

Updates

posted an update —

I have completed the update of the prototype to Iteration two with improved performance and reliability as planned for the weekend. The app now uses the bodyPix model version 2. We can focus on the video and the submission for the remainder of the hackathon.

Log in or sign up for Devpost to join the conversation.

posted an update —

I am working on the second iteration of this prototype. Iteration one was a proof of concept made with p5.js, ml5.js (bodyPix model version 1). The performance and reliability are in need of improvement. Iteration 2 gets rid of all frameworks except for Tensorflow and switches to bodyPix model version 2. I hope this will solve the current problems.

Log in or sign up for Devpost to join the conversation.