the rgbd device in skeletonRetriever and lineDetector is now closed after retrieving the camera parameters to save bandwidth (otherwise they keep receiving depth images );
depth quantization for the realsense is enabled to allow compressing depth images using zlib;
the initial guess for the line pose estimate in lineDetector is now set to false by default and parameterized. This is to avoid a bad initial guess to affect detrimentally next estimates.
This PR includes the following for the TUG demo:
skeletonRetrieverandlineDetectoris now closed after retrieving the camera parameters to save bandwidth (otherwise they keep receiving depth images );zlib;lineDetectoris now set tofalseby default and parameterized. This is to avoid a bad initial guess to affect detrimentally next estimates.