Global Hierarchical Attention for 3D Point Cloud Analysis

TL;DR

This paper introduces Global Hierarchical Attention (GHA), a scalable attention mechanism for 3D point cloud analysis that improves performance across multiple tasks by efficiently capturing global and local spatial relationships.

Contribution

The paper presents GHA, a novel attention method with linear complexity that enhances existing 3D point cloud networks by effectively modeling global connectivity.

Findings

GHA increases mIoU by 1.7% on ScanNet semantic segmentation.

GHA improves CenterPoint 3D detection by 0.5% mAP on nuScenes.

GHA boosts 3DETR performance by 2.1% mAP25 on ScanNet.

Abstract

We propose a new attention mechanism, called Global Hierarchical Attention (GHA), for 3D point cloud analysis. GHA approximates the regular global dot-product attention via a series of coarsening and interpolation operations over multiple hierarchy levels. The advantage of GHA is two-fold. First, it has linear complexity with respect to the number of points, enabling the processing of large point clouds. Second, GHA inherently possesses the inductive bias to focus on spatially close points, while retaining the global connectivity among all points. Combined with a feedforward network, GHA can be inserted into many existing network architectures. We experiment with multiple baseline networks and show that adding GHA consistently improves performance across different tasks and datasets. For the task of semantic segmentation, GHA gives a +1.7% mIoU increase to the MinkowskiEngine baseline on ScanNet. For the 3D object detection task, GHA improves the CenterPoint baseline by +0.5% mAP on the nuScenes dataset, and the 3DETR baseline by +2.1% mAP25 and +1.5% mAP50 on ScanNet.