Is there a way to offline skeleton acquisition? RRS feed

  • Question

  • I learned that Kinect provide skeleton from depth image, so I wonder it's there a way that I record depth image and do some modify(filter or other things), and then input it into kinect to get the skeleton?
    Wednesday, November 18, 2015 8:47 AM

All replies

  • Kinect for Windows 2 API does not expose the skeleton tracking algorithm explicitly to allow the scenario you described. It is available only as one of the "feeds" from the sensor.

    You need to use a 3rd party skeleton tracking algorithm for this.

    • Proposed as answer by Jan Marcincin Wednesday, November 18, 2015 9:10 AM
    Wednesday, November 18, 2015 9:10 AM
  • Thanks for you reply, do you know which 3rd party can do this? I search for it for 2 days, but can't find one
    Thursday, November 19, 2015 2:37 AM
  • OpenNI has some "simple" skeleton tracking implemented - you can get an inspiration from it. 

    I am also not aware of any library with skeleton tracking on the level of Kinect runtime.

    This is usually a topic for academia - you might get an access to some research papers covering this topic but you will most likely have to write the algorithm yourself. 

    Thursday, November 19, 2015 7:08 AM