Rating is available when the video has been rented.
This feature is not available right now. Please try again later.
Published on Apr 30, 2009
This is a very rough test of my present "face recognition" s/w. I've just channeled a couple of 60-sec clips through the prog running on a slow old x86. It manages to process at around 20 fps, and has an error rate of around 3%. But when it recognises a pattern in the grass or on the ear of a cat as a human face, the error tends to be spectacular. I'll activate a particle filter and/or Kalman filter on the motions so the tracking will be more reliable and smoother. But that's a project for next week.