Abstract: The basic components of a point-based 3D object detector is set abstraction (SA) layer, which down-samples points for better efficiency and enlarges receptive fields. However, existing SA ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results