DV-Det: Efficient 3D Point Cloud Object Detection with Dynamic Voxelization

TL;DR

DV-Det introduces a fast and accurate 3D point cloud object detection framework that uses dynamic voxelization to directly process raw point clouds, avoiding costly transformations and complex learning modules.

Contribution

The paper presents dynamic voxelization for real-time 3D detection, reducing reliance on MLPs and conventional convolutions, and offers CUDA-optimized implementations for speed.

Findings

75 FPS on KITTI dataset

25 FPS on Waymo dataset

Achieves competitive accuracy with high efficiency

Abstract

In this work, we propose a novel two-stage framework for the efficient 3D point cloud object detection. Instead of transforming point clouds into 2D bird eye view projections, we parse the raw point cloud data directly in the 3D space yet achieve impressive efficiency and accuracy. To achieve this goal, we propose dynamic voxelization, a method that voxellizes points at local scale on-the-fly. By doing so, we preserve the point cloud geometry with 3D voxels, and therefore waive the dependence on expensive MLPs to learn from point coordinates. On the other hand, we inherently still follow the same processing pattern as point-wise methods (e.g., PointNet) and no longer suffer from the quantization issue like conventional convolutions. For further speed optimization, we propose the grid-based downsampling and voxelization method, and provide different CUDA implementations to accommodate to the discrepant requirements during training and inference phases. We highlight our efficiency on KITTI 3D object detection dataset with 75 FPS and on Waymo Open dataset with 25 FPS inference speed with satisfactory accuracy.