Abstract: Applying convolution methods to domains that lack regular underlying structures is a challenging task for 3D vision. Existing methods require the manual design of feature representations ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results