0

I want to get a fast and good approximated segmentation of my object within its detection box. How could I do that? I tried background subtraction algorithms and grab cut but they dont look that promising.

Kostas
  • 270
  • 1
  • 3
  • 10
  • 1
    Are you using a single web cam? What do you exactly want? What is your setup? Can you make any assumptions regarding the scene? Please be a little bit more specific. – Markus Mayr May 08 '15 at 10:22
  • Its an indoor application of a person walking in a room. Im using Kinect Sensor; It is mounted around 2m from the ground. Im using deformable part models for detecting the person. Thus I have the main detection box plus the smaller detection parts which where used to get the main detection box. I also have the detection score; – Kostas May 08 '15 at 12:54
  • @Kostas can you post an example image + detection result + what you are expecting to get? why were you not satisfied with GrabCut? do you have an additional depth information? – Shai May 10 '15 at 09:17
  • have you considered [this answer](http://stackoverflow.com/a/27938800/1714410)? – Shai May 10 '15 at 09:19

0 Answers0