Deep Parametric Continuous Convolutional Neural Networks

TL;DR

This paper introduces Parametric Continuous Convolution, a learnable operator that extends CNNs to non-grid data, enabling improved performance in point cloud segmentation and lidar motion estimation.

Contribution

It proposes a novel continuous convolution operator that generalizes traditional CNNs to arbitrary data structures with learnable kernel functions.

Findings

Significant improvement in point cloud segmentation accuracy.

Enhanced lidar motion estimation performance.

Applicable to diverse non-grid structured data.

Abstract

Standard convolutional neural networks assume a grid structured input is available and exploit discrete convolutions as their fundamental building blocks. This limits their applicability to many real-world applications. In this paper we propose Parametric Continuous Convolution, a new learnable operator that operates over non-grid structured data. The key idea is to exploit parameterized kernel functions that span the full continuous vector space. This generalization allows us to learn over arbitrary data structures as long as their support relationship is computable. Our experiments show significant improvement over the state-of-the-art in point cloud segmentation of indoor and outdoor scenes, and lidar motion estimation of driving scenes.