PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space

PointNet pioneered deep learning on point sets, but by design it does not capture local structures induced by the metric space points live in, limiting its ability to recognize fine-grained patterns and generalize to complex scenes. This work introduces PointNet++, a hierarchical neural network that applies PointNet recursively on a nested partitioning of the input point set, exploiting metric space distances to learn local features with increasing contextual scales.

Observing that point sets are usually sampled with varying densities, which greatly decreases performance for networks trained on uniform densities, the authors also propose novel set learning layers that adaptively combine features from multiple scales. Experiments show that PointNet++ learns deep point set features efficiently and robustly, obtaining results significantly better than the state of the art on challenging benchmarks of 3D point clouds.