lysdexic-audio/jweb-pose-landmarker

A self contained example demonstrating how to use Mediapipe Pose Landmarker with Max's `jweb`

lysdexic-audio 👋

I make sound and interaction design & composition tools. MaxMSP, C++, JS, Python etc.

For the last few years I've been full-time as a Senior C++ Audio Software Engineer at Aimi.fm - which takes most of my time. I continue to create and update repos here from time to time, please create issues for bug reports and feature requests.

🏢 Primary / work account: https://github.com/jcurtis-cc

---

Google Mediapipe projects

Max - jweb

• Pose (body) : https://github.com/lysdexic-audio/jweb-pose-landmarker
• Hand tracking: https://github.com/lysdexic-audio/jweb-hands-landmarker
• Hand gestures: https://github.com/lysdexic-audio/jweb-hands-gesture-recognizer
• Face tracking: https://github.com/lysdexic-audio/jweb-facemesh
• Face gestures: https://github.com/lysdexic-audio/jweb-face-landmarker
• Object detection: https://github.com/lysdexic-audio/jweb-object-detection

Max - node for max

• Handtracking : https://github.com/lysdexic-audio/n4m-handpose
• Facetracking : https://github.com/lysdexic-audio/n4m-facemesh

jweb-pose-landmarker's Issues

Input from other jit objects?

[sorry for cross post of questions with cycling 74 forum]
Thanks Lysdexic. This is fantastic work!

I'm interested in feeding in an image/matrix from a Kinect or other IR camera (the goal is being able to send a live image of a person in a dark room that is hopefully readable enough by Pose to make out a body and limbs). For matrix data from the Kinect, I'm using jit.freenect.grab.
Is there a way to feed another jit object directly into the jweb-pose-landmarker object to be analyzed rather than an image from media device/camera?

I'm guessing the js code for jweb-pose-landmarker would have to be changed for a different input, but I don't know where to start.

Thanks again!
Jacob

Add still image option

Request from C74 forum

New model is very jittery

Smooth the values with a buffer perhaps