Papers › DSVT: Dynamic Sparse Voxel Transformer with Rotated Sets

DSVT: Dynamic Sparse Voxel Transformer with Rotated Sets

15 Jan 2023CVPR 2023 1arXiv:2301.06051archive 2025-07-28

Haiyang Wang, Chen Shi, Shaoshuai Shi, Meng Lei, Sen Wang, Di He, Bernt Schiele, LiWei Wang

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}.

PaperPDFConference PDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

haiyang-w/dsvt officialmentioned in papermentioned on GitHubpytorchApache-2.0 report
open-mmlab/OpenPCDet officialmentioned in papermentioned on GitHubpytorchApache-2.0 report
happytianhao/tade mentioned on GitHubpytorchApache-2.0 report
open-mmlab/mmdetection3d mentioned on GitHubpytorchApache-2.0 report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

3D Object DetectionObject Detectionobject-detection

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
3D Object Detection Waymo Open Dataset DSVT mAPH/L2 72.1 #4 of 8 Archive leaderboard report
3D Object Detection nuScenes DSVT NDS 0.73 #47 of 372 Archive leaderboard report
3D Object Detection nuScenes DSVT mAAE 0.14 #47 of 372 Archive leaderboard report
3D Object Detection nuScenes DSVT mAOE 0.30 #47 of 372 Archive leaderboard report
3D Object Detection nuScenes DSVT mASE 0.23 #47 of 372 Archive leaderboard report
3D Object Detection nuScenes DSVT mATE 0.25 #47 of 372 Archive leaderboard report
3D Object Detection nuScenes DSVT mAVE 0.25 #47 of 372 Archive leaderboard report
3D Object Detection nuScenes LiDAR only DSVT NDS 72.7 #2 of 7 Archive leaderboard report
3D Object Detection nuScenes LiDAR only DSVT NDS (val) 71.1 #2 of 7 Archive leaderboard report
3D Object Detection nuScenes LiDAR only DSVT mAP 68.4 #2 of 7 Archive leaderboard report
3D Object Detection nuScenes LiDAR only DSVT mAP (val) 66.4 #2 of 7 Archive leaderboard report
3D Object Detection waymo cyclist DSVT(val) APH/L2 78.0 #1 of 7 Archive leaderboard report
3D Object Detection waymo pedestrian DSVT(val) APH/L2 76.4 #1 of 7 Archive leaderboard report
3D Object Detection waymo vehicle DSVT(val) APH/L2 74.1 #2 of 8 Archive leaderboard report
3D Object Detection waymo vehicle DSVT(val) L1 mAP 82.1 #2 of 8 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Absolute Position EncodingsAdamAttentionBPEDense ConnectionsDropoutLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSPEEDSoftmaxTransformer

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections