Explaining Hierarchical Features in Dynamic Point Cloud Processing

TL;DR

This paper investigates how hierarchical features are learned and represented in deep learning models for dynamic point cloud processing, aiming to clarify the roles of different network components in feature learning.

Contribution

It provides an analysis of hierarchical feature learning in deep networks for point cloud prediction, highlighting how network structure influences learned features and their significance.

Findings

Hierarchical features vary across network stages.

Network components significantly impact feature learning.

Insights aid in designing better point cloud processing models.

Abstract

This paper aims at bringing some light and understanding to the field of deep learning for dynamic point cloud processing. Specifically, we focus on the hierarchical features learning aspect, with the ultimate goal of understanding which features are learned at the different stages of the process and what their meaning is. Last, we bring clarity on how hierarchical components of the network affect the learned features and their importance for a successful learning model. This study is conducted for point cloud prediction tasks, useful for predicting coding applications.