Hi David, fyio, to even better improve the localization of the face and to filter out false positives I'd often use face detector in the combination with the legs detector, e.g. http://tum-ros-pkg.svn.sourceforge.net/viewvc/tum-ros-pkg/perception/people_detector_node/. I did not check as of lately but I am pretty sure that Caroline also has one in the people stack. cheers, D. On Sun, Aug 1, 2010 at 6:49 PM, Caroline Pantofaru wrote: > Hi David > Right now the face detector uses the median disparity within the face > bounding box to check that the face is: > - less than 8 meters from the camera > - has a width greater than 10cm > - has a width less than 50cm > Those limits are pretty arbitrary, I can expose them as parameters in the > next release. Please create a ticket and assign it to me so that I don't > forget. > You might want to check the disparities coming from your cameras. The limits > I've set on the face size are very loose, so I'm surprised that you aren't > getting any acceptable faces. > -Caroline > On Sat, Jul 31, 2010 at 7:13 PM, David Lu!! wrote: >> >> Hey all, >> I'm running the face detector from the newly released people stack. It >> finds faces well, but none with good depth and size. What qualifies as good >> depth and size? >> Thanks, >> David!! >> _______________________________________________ >> ros-users mailing list >> ros-users@code.ros.org >> https://code.ros.org/mailman/listinfo/ros-users >> > > > _______________________________________________ > ros-users mailing list > ros-users@code.ros.org > https://code.ros.org/mailman/listinfo/ros-users > > -- MSc. Dejan Pangercic PhD Student/Researcher Intelligent Autonomous Systems Group Technische Universität München Telephone: +49 (89) 289-26908 E-Mail: dejan.pangercic@cs.tum.edu WWW: http://ias.cs.tum.edu/people/pangercic