DSVT: Dynamic Sparse Voxel Transformer with Rotated Sets

TL;DR

DSVT introduces a novel sparse voxel transformer backbone for 3D perception that efficiently models long-range relationships in sparse point clouds, achieving state-of-the-art performance with real-time inference speed.

Contribution

The paper proposes Dynamic Sparse Voxel Transformer (DSVT), a new transformer-based method with dynamic sparse window attention and rotated set partitioning for efficient 3D perception.

Findings

Achieves state-of-the-art results on 3D perception tasks.

Supports real-time inference at 27Hz with TensorRT.

Effective encoding of geometric information without custom CUDA operations.

Abstract

Designing an efficient yet deployment-friendly 3D backbone to handle sparse point clouds is a fundamental problem in 3D perception. Compared with the customized sparse convolution, the attention mechanism in Transformers is more appropriate for flexibly modeling long-range relationships and is easier to be deployed in real-world applications. However, due to the sparse characteristics of point clouds, it is non-trivial to apply a standard transformer on sparse points. In this paper, we present Dynamic Sparse Voxel Transformer (DSVT), a single-stride window-based voxel Transformer backbone for outdoor 3D perception. In order to efficiently process sparse points in parallel, we propose Dynamic Sparse Window Attention, which partitions a series of local regions in each window according to its sparsity and then computes the features of all regions in a fully parallel manner. To allow the cross-set connection, we design a rotated set partitioning strategy that alternates between two partitioning configurations in consecutive self-attention layers. To support effective downsampling and better encode geometric information, we also propose an attention-style 3D pooling module on sparse points, which is powerful and deployment-friendly without utilizing any customized CUDA operations. Our model achieves state-of-the-art performance with a broad range of 3D perception tasks. More importantly, DSVT can be easily deployed by TensorRT with real-time inference speed (27Hz). Code will be available at \url{https://github.com/Haiyang-W/DSVT}.