CameraPoseFinder - Save and Load Capabilities RRS feed

  • Question

  • After some digging around this weekend, I found that the CameraPoseFinder interface has a method for saving and loading the database of CameraPoses.

    I've tried using the save method, but there is no information as to what format the information is in, etc.  It appears to be a binary file and that the first "frame" of pose information starts at 85f4(hex).  Each frame appears to be 30(hex) bytes long.  Other than byte 85f0 indicating the number of frames in the db, I can't seem to discern anything else about the file/format.

    Does any one have any other details/guidance on how to utilize the save/load functionality of the CameraPoseFinder.


    Monday, January 20, 2014 6:06 PM


All replies

  • The data is only supported for use with the LoadCameraPoseFinderDatabase function. It is undocumented as the data is subject to change without notice. Any modification or change the data is not supported.

    What exactly are you looking to do?

    Carmine Sirignano - MSFT

    Monday, January 20, 2014 11:55 PM
  • Thanks Carmine.

    I'm working on my thesis and was looking to quantify the pose estimation vs the actual pose of the Kinect.  To that end, I was hoping to export the X,Y,Z location of the Pose Estimation and compare it to a high confidence X,Y,Z value.  Does that make sense?

    (The follow on would be to look at using the external pose information in addition to the current pose estimation to see if it helps make better Fusion models.)

    Tuesday, January 21, 2014 12:13 AM