none
Getting head rotation from nearby points? RRS feed

  • General discussion

  • I know I could try instead of asking, but has anyone here tried getting the head rotation (as in a vector that gives the direction of a nose ;)) by sampling points near the head point position and calculating a normal? If so does that work well enough?

    • Changed type Eddy Escardo-Raffo [MSFT] Wednesday, September 7, 2011 1:44 AM Looks like more of a discussion question than something that will get a simple straightforward answer
    Tuesday, September 6, 2011 2:50 PM

All replies

  • You are talking about head pose estimation. It seems like using a normal from the nose, mouth or center of the forehead could work in concept, but I'm wondering what kind of approach you will use to know where those facial features are?

    There were at least two papers dealing explicitly with head pose estimation at the 2011 IEEE Conference on Computer Vision and Pattern Recognition (CVPR, see http://www.cvpapers.com/cvpr2011.html). You should know CVPR because the Microsoft Research Kinect team was awarded Best Paper 2011 for their great work!

    Thursday, September 8, 2011 9:07 AM