Abstract: Existing methods for place recognition in 3D point clouds either ignore partial structure information by converting 3D scans to 2D images or construct constrained bag-of-words (BoW) ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback