Gazebo / URDF human model (with joints, not an actor)

Hello, I want to use deep learning with depth data from a human body in different poses. The easiest way to obtain a big amount of labelled data is to simulate this human body and obtain labelled depth images.

For this I need a human model either directly modelled in Gazebo SDF or a URDF model. I've been looking for some kind of mannequin model (like the typical wooden mannequin) but I haven't found anything close.

The closest were this question, non moveable and this other question, animated model.

So I'm asking if anyone has something like what I'm looking for. Otherwise, I'll start implementing it, but as it will be quite a lot of work (specially because I'd like to vary the size and volume of links to simulate different sized humans) any help is welcome.

Edit: I may start by modelling from some 3D mannequin model already, like this one.

Asked by awesomebytes on 2017-03-30 06:54:34 UTC

Comments

Hello! Did you finally manage to create a human model and got it working?

Asked by rayvburn on 2018-12-09 07:25:22 UTC

Answers

Is this what you are looking for?

Asked by nkoenig on 2017-03-30 10:07:22 UTC

Comments

No, I want a model where I can move all the joints, not an actor with predefined motions. So like a robot model but of a human. I don't need it to be that realistic as I want to use it with depth images so the mannequin approach may work. Thank you for your answer tho.

Asked by awesomebytes on 2017-04-11 20:34:01 UTC

I think you'll have to create your own mannequin, unless someone else chimes in with a model.

Asked by nkoenig on 2018-12-11 18:02:23 UTC