Body tracking without special cameras

Homepage Forums Body tracking Body tracking without special cameras

Viewing 4 posts - 1 through 4 (of 4 total)
  • Author
    Posts
  • #15844
    ERIC LIMAWHITHIL
    Participant

    After seeing some interesting projects like
    PoseNet [https://github.com/tensorflow/tfjs-models/tree/master/posenet]
    And how they can provide real time skeletal model of a user just like a Kinect does I would like to discuss the feasibility of such functions on Driver4VR.

    Would it work?

    #15845
    ERIC LIMAWHITHIL
    Participant

    Here’s an idea:
    Intead of using the PC to compute the skeletal model and waste precious resources that could be used by, well, the game itself.
    We could use a Mobile companion App.

    This would keep the hard work on another CPU, so the PC just need to have a “server” to receive in plain text or whatever. The description of the skeletal model and do whatever black magic it’s doing with a Kinect camera right now 🤷‍♂️👍

    #15846
    ERIC LIMAWHITHIL
    Participant

    Also. There’s no IR Grid like it is used on Kinect, so there would be no interference when using 2 or more cameras.

    And this way you could easily add more precision to it.

    So here’s a low cost scenario that could really be great:

    * 1 PC with a webcam;
    * 1 Mobile Phone;
    * DRIVER4VR in tVR mode (because you may have just 1 phone and no headsets, so phoneVR would not be possible);

    And it would just work great with very low to no cost at all 😀

    #15847
    ERIC LIMAWHITHIL
    Participant

    Also, instead of developing all kinds of support for specific tracking hardware and techniques.

    There could be just a simple JSON API for skeletal and position tracking.
    And then just a tracking server running on port XXXX and then let the community experiment with ideas and contribute more directly.

Viewing 4 posts - 1 through 4 (of 4 total)
  • You must be logged in to reply to this topic.