Inferred Joint Algorithm RRS feed

  • Question

  • Hi,

    i am a little bit curious about how kinect estimate joint position when it is INFERRED. What algorithm it uses?

    Thursday, April 5, 2012 7:54 AM

All replies

  • There are several reasons that the tracking state for a joint may be reported as inferred. Generally, it means that the algorithm's confidence in estimating the position did not meet a minimum threshold. This may be due to the joint being hidden from view (e.g., by another body part) or outside the camera's field of view (too high or too low, or off to the side). It may also result from too much ambiguity in the depth data, causing the algorithm to not be able to definitively choose between two or more possible positions for the joint.

    If you're interested in how the algorithm works, this paper by Microsoft Research explains it:

    K4W Dev

    Thursday, April 5, 2012 6:00 PM
  • Hi!

    I am using a Kinect(v2) to capture ankle joint position data during walking. Sometime the ankle joint must be getting occluded for the contralateral limb, but I am getting a time series (having some numerical values in .csv file format) for both ankle joints.

    Can anyone tell me how to identify the inferred body points (the occluded positions) from the time series?

    Thanks in advance.

    • Edited by SaikatNITRKL Saturday, February 2, 2019 12:22 PM
    Saturday, February 2, 2019 12:18 PM