TL;DR本文提出了一种新型的图卷积神经网络,名为差分图卷积 (diffConv),通过学习掩码式注意机制来适应空间变化和密度膨胀的邻域,相比现有方法在 3D 形状分类和场景理解任务上获得了更好的性能。
Abstract
Standard spatial convolutions assume input data with a regular neighborhood
structure. Existing methods typically generalize convolution to the irregular
point cloud domain by fixing a regular "view" through e.g. a fixed neighborhood
size, where the convolution kernel size remains the