Abstract: Convolution on 3D point clouds is widely researched yet far from perfect in geometric deep learning. The traditional wisdom of convolution characterises feature correspondences ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback