none
why microsoft gave only 20 Joints with Kinect ? why not more than 20 ? RRS feed

  • Question

  • why microsoft gave only 20 Joints with Kinect ?

    what is reason for choosing only 20 they could have come up with more 

    Friday, July 12, 2013 1:32 PM

Answers

  • Skeleton tracking relies on the Depth sensor which has a low resolution (640x480).

    This resolution can be enough to detect hand detailed (fingers) at close range.
    But the primary use of Kinect is for full or half skeleton recognition, which mean user should be at least 2 meters for a full detection.

    At this distance and with this resolution, there is not enough information (or confidence) for detecting fingers and guaranting proper detection for joints other than those 20.

    One big added-value from Microsoft Kinect SDK is allowing developers to use raw sensor stream in order to create custom detection algorithm. Few open-source algorithm are available on the web (have a look at OpenCV for alternative detection and tracking algorithm)


    Vincent Guigui Innovative Technologies Expert at OCTO Technology MVP Kinect

    Saturday, July 13, 2013 9:17 PM

All replies

  • The 20 joints represent the major joints of the human body to articulate motion/animation. Which other joints are you referring to?

    http://msdn.microsoft.com/en-us/library/jj131025.aspx#ID4EHE


    Carmine Si - MSFT

    Friday, July 12, 2013 5:20 PM
  • That's true  that those are  min 20 joints required to articulate motion  of a human body.But there are around 200-300 joints among  bones ,for example on a single  palm we have around more than  10 joints.I think it is the limitation of sensor or left to developer to play around and look for more.

    Saturday, July 13, 2013 6:12 AM
  • Skeleton tracking relies on the Depth sensor which has a low resolution (640x480).

    This resolution can be enough to detect hand detailed (fingers) at close range.
    But the primary use of Kinect is for full or half skeleton recognition, which mean user should be at least 2 meters for a full detection.

    At this distance and with this resolution, there is not enough information (or confidence) for detecting fingers and guaranting proper detection for joints other than those 20.

    One big added-value from Microsoft Kinect SDK is allowing developers to use raw sensor stream in order to create custom detection algorithm. Few open-source algorithm are available on the web (have a look at OpenCV for alternative detection and tracking algorithm)


    Vincent Guigui Innovative Technologies Expert at OCTO Technology MVP Kinect

    Saturday, July 13, 2013 9:17 PM