Hierarchical Point-Edge Interaction Network for Point Cloud Semantic Segmentation

TL;DR

This paper introduces a hierarchical point-edge interaction network that leverages semantic relations between points and edges for improved 3D point cloud semantic segmentation, integrating hierarchical graph features and edge-based semantic consistency predictions.

Contribution

It proposes a novel hierarchical graph framework with an edge branch that enhances point features and segmentation accuracy through semantic consistency modeling.

Findings

Effective on multiple 3D semantic datasets

Improves segmentation accuracy with hierarchical edge features

Enhances local contextual message passing

Abstract

We achieve 3D semantic scene labeling by exploring semantic relation between each point and its contextual neighbors through edges. Besides an encoder-decoder branch for predicting point labels, we construct an edge branch to hierarchically integrate point features and generate edge features. To incorporate point features in the edge branch, we establish a hierarchical graph framework, where the graph is initialized from a coarse layer and gradually enriched along the point decoding process. For each edge in the final graph, we predict a label to indicate the semantic consistency of the two connected points to enhance point prediction. At different layers, edge features are also fed into the corresponding point module to integrate contextual information for message passing enhancement in local regions. The two branches interact with each other and cooperate in segmentation. Decent experimental results on several 3D semantic labeling datasets demonstrate the effectiveness of our work.