Rethinking PointNet Embedding for Faster and Compact Model

TL;DR

This paper proposes replacing PointNet's embedding function with Gaussian kernels to create a faster, more compact model that maintains accuracy while significantly reducing computational costs for point cloud processing.

Contribution

It introduces a novel Gaussian kernel-based embedding method for PointNet, achieving comparable performance with up to 92% fewer floating-point operations.

Findings

Achieves similar accuracy to baseline PointNet models.

Reduces computational cost by up to 92%.

Maintains universal approximation capabilities.

Abstract

PointNet, which is the widely used point-wise embedding method and known as a universal approximator for continuous set functions, can process one million points per second. Nevertheless, real-time inference for the recent development of high-performing sensors is still challenging with existing neural network-based methods, including PointNet. In ordinary cases, the embedding function of PointNet behaves like a soft-indicator function that is activated when the input points exist in a certain local region of the input space. Leveraging this property, we reduce the computational costs of point-wise embedding by replacing the embedding function of PointNet with the soft-indicator function by Gaussian kernels. Moreover, we show that the Gaussian kernels also satisfy the universal approximation theorem that PointNet satisfies. In experiments, we verify that our model using the Gaussian kernels achieves comparable results to baseline methods, but with much fewer floating-point operations per sample up to 92% reduction from PointNet.