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

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.

Accepted by International Journal of Computer Vision (IJCV), code is available at https://github.com/open-mmlab/OpenPCDet

References in corpus (8)

• 3D-CVF: Generating Joint Camera and LiDAR Features Using Cross-View Spatial Feature Fusion for 3D Object Detection
• Deep Continuous Fusion for Multi-Sensor 3D Object Detection
• Submanifold Sparse Convolutional Networks
• HDNET: Exploiting HD Maps for 3D Object Detection
• Point-GNN: Graph Neural Network for 3D Object Detection in a Point Cloud
• Voxel-FPN: multi-scale voxel feature aggregation in 3D object detection from point clouds
• Pillar-based Object Detection for Autonomous Driving
• A Closer Look at Local Aggregation Operators in Point Cloud Analysis

Cited by in corpus (8)

• Pyramid R-CNN: Towards Better Performance and Adaptability for 3D Object Detection
• ST3D: Self-training for Unsupervised Domain Adaptation on 3D Object Detection
• Joint stereo 3D object detection and implicit surface reconstruction
• Structure Information is the Key: Self-Attention RoI Feature Extractor in 3D Object Detection
• Improved Pillar with Fine-grained Feature for 3D Object Detection
• Dynamic Convolution for 3D Point Cloud Instance Segmentation
• Investigating the Impact of Multi-LiDAR Placement on Object Detection for Autonomous Driving
• Multi Voxel-Point Neurons Convolution (MVPConv) for Fast and Accurate 3D Deep Learning