SPLATNet: Sparse Lattice Networks for Point Cloud Processing

TL;DR

SPLATNet introduces a sparse lattice-based neural network architecture for efficient point cloud processing, enabling hierarchical, spatially-aware features and joint 2D-3D reasoning, outperforming existing methods.

Contribution

The paper proposes a novel sparse lattice convolutional network architecture that efficiently processes point clouds with hierarchical and spatially-aware features.

Findings

Outperforms state-of-the-art 3D segmentation methods

Efficiently handles large point clouds with sparse convolutions

Supports joint 2D-3D reasoning in a unified framework

Abstract

We present a network architecture for processing point clouds that directly operates on a collection of points represented as a sparse set of samples in a high-dimensional lattice. Naively applying convolutions on this lattice scales poorly, both in terms of memory and computational cost, as the size of the lattice increases. Instead, our network uses sparse bilateral convolutional layers as building blocks. These layers maintain efficiency by using indexing structures to apply convolutions only on occupied parts of the lattice, and allow flexible specifications of the lattice structure enabling hierarchical and spatially-aware feature learning, as well as joint 2D-3D reasoning. Both point-based and image-based representations can be easily incorporated in a network with such layers and the resulting model can be trained in an end-to-end manner. We present results on 3D segmentation tasks where our approach outperforms existing state-of-the-art techniques.