Turn yourself into an SVG with Pose Animator, a deep learning recognition tool

May 11th
(Shan Huang)

Shan Huang, a creative technologist at Google, recently posted a side project on Twitter showing an avatar mimic her movements in real time.

You can do it too. The demo, called Pose Animator, reacts in real time to your movements. It's easy to set up -- just open it in a browser, enable your webcam, and start moving. Huang says that it will work on Chrome and iOS Safari (it's a little slow on Brave).

Huang explains how it works in her repo:

In Pose Animator, the surface is defined by the 2D vector paths in the input SVG files. For the bone structure, Pose Animator provides a predefined rig (bone hierarchy) representation, designed based on the keypoints from PoseNet and FaceMesh. This bone structure’s initial pose is specified in the input SVG file, along with the character illustration, while the real time bone positions are updated by the recognition result from ML models.

You can also upload your own SVG files, and animate yourself however you want.