Object recognition capture unable to find Pose

edit

I have an Ubuntu 12.10 with ROS groovy configuration. I use a kinect on a USB2.0 port.

I would like to use the object_recognition package but I don't succeed in make the capture tool work. I succeed capturing my textured_plane but impossible to have a 3D position displayed when I track the features.

Here is my command line for capturing my textured plane :

rosrun object_recognition_capture orb_template -o my_textured_plane --res sxga --fps 15

But when I try to see the position tracking, no position is outputed in the video

rosrun object_recognition_capture orb_track --track_directory my_textured_plane --res sxga --fps 15

But, if I show the matches, the features detection seems to work quite well.

So I try some different parameters (more feature points, less threshold ...) but the position never appears.

Does anyone have a clue on why the 3D position is not display while the features are correctly detected?

PS : I also try rosrun object_recognition_capture capture -i my_textured_plane --seg_z_min 0.01 -o silk.bag --preview --res sxga --fps 15 but no object is detected.