PV-RCNN++: Point-Voxel Feature Set Abstraction With Local Vector Representation for 3D Object Detection

TL;DR

This paper introduces PV-RCNN++: an advanced 3D object detection framework that combines point and voxel features, significantly improving speed and accuracy for autonomous driving applications.

Contribution

It presents PV-RCNN++, a novel 3D detection method with efficient keypoint sampling and local feature aggregation, outperforming previous models in speed and accuracy.

Findings

PV-RCNN++ is about 3 times faster than PV-RCNN.

Achieves state-of-the-art performance on Waymo Open Dataset.

Operates at 10 FPS within a 150m range.

Abstract

3D object detection is receiving increasing attention from both industry and academia thanks to its wide applications in various fields. In this paper, we propose Point-Voxel Region-based Convolution Neural Networks (PV-RCNNs) for 3D object detection on point clouds. First, we propose a novel 3D detector, PV-RCNN, which boosts the 3D detection performance by deeply integrating the feature learning of both point-based set abstraction and voxel-based sparse convolution through two novel steps, i.e., the voxel-to-keypoint scene encoding and the keypoint-to-grid RoI feature abstraction. Second, we propose an advanced framework, PV-RCNN++, for more efficient and accurate 3D object detection. It consists of two major improvements: sectorized proposal-centric sampling for efficiently producing more representative keypoints, and VectorPool aggregation for better aggregating local point features with much less resource consumption. With these two strategies, our PV-RCNN++ is about $3\times$ faster than PV-RCNN, while also achieving better performance. The experiments demonstrate that our proposed PV-RCNN++ framework achieves state-of-the-art 3D detection performance on the large-scale and highly-competitive Waymo Open Dataset with 10 FPS inference speed on the detection range of 150m * 150m.