Google's Move Mirror Tracks Users' Movements and Compares it to Images

 - Jul 21, 2018
References: engadget & cnet
Google's AI platform has already allowed users to match themselves up with artwork, but the company's latest Move Mirror system does more than just compare faces to priceless pieces of artwork. Release in a blog post on July 19, 2018, this new feature serves as a machine learning experiment that will hopefully inspire coders and makers to use Google tech in interesting ways. To create this AI feature, Google utilized two open-source programs known as PoseNet, which detects body poses, and TensorFlow.js, a library for in-browser machine learning framework.

Like Google Arts and Culture, Move Mirror utilizes a camera to track a user and them taps into a database to pull up an image similar to what the user is doing. None of the user's images are sent to any Google servers, with Move Mirror handling all the image recognition locally within the browser.

Image Credit: Google