{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/dsvt-dynamic-sparse-voxel-transformer-with","title":"DSVT: Dynamic Sparse Voxel Transformer with Rotated Sets","arxiv_id":"2301.06051","date":"2023-01-15","proceeding":"CVPR 2023 1","authors":["Haiyang Wang","Chen Shi","Shaoshuai Shi","Meng Lei","Sen Wang","Di He","Bernt Schiele","LiWei Wang"],"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}.","url_abs":"https://arxiv.org/abs/2301.06051v2","url_pdf":"https://arxiv.org/pdf/2301.06051v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"dsvt-dynamic-sparse-voxel-transformer-with","repo_url":"https://github.com/haiyang-w/dsvt","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"dsvt-dynamic-sparse-voxel-transformer-with","repo_url":"https://github.com/open-mmlab/OpenPCDet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"dsvt-dynamic-sparse-voxel-transformer-with","repo_url":"https://github.com/happytianhao/tade","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"dsvt-dynamic-sparse-voxel-transformer-with","repo_url":"https://github.com/open-mmlab/mmdetection3d","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"3d-object-detection","task_name":"3D Object Detection"},{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":"object-detection-1","task_name":"object-detection"}],"methods":[{"method_slug":"absolute-position-encodings","method_name":"Absolute Position Encodings"},{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"label-smoothing","method_name":"Label Smoothing"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"position-wise-feed-forward-layer","method_name":"Position-Wise Feed-Forward Layer"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"speed","method_name":"SPEED"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"transformer","method_name":"Transformer"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-object-detection-on-waymo-open-dataset","task":"3D Object Detection","dataset":"Waymo Open Dataset","model":"DSVT","rank_in_archive_order":4,"of":8,"metrics":{"mAPH/L2":"72.1"},"uses_additional_data":false},{"leaderboard":"/sota/3d-object-detection-on-nuscenes","task":"3D Object Detection","dataset":"nuScenes","model":"DSVT","rank_in_archive_order":47,"of":372,"metrics":{"NDS":"0.73","mAAE":"0.14","mAOE":"0.30","mASE":"0.23","mATE":"0.25","mAVE":"0.25"},"uses_additional_data":false},{"leaderboard":"/sota/3d-object-detection-on-nuscenes-lidar-only","task":"3D Object Detection","dataset":"nuScenes LiDAR only","model":"DSVT","rank_in_archive_order":2,"of":7,"metrics":{"NDS":"72.7","NDS (val)":"71.1","mAP":"68.4","mAP (val)":"66.4"},"uses_additional_data":false},{"leaderboard":"/sota/3d-object-detection-on-waymo-cyclist","task":"3D Object Detection","dataset":"waymo cyclist","model":"DSVT(val)","rank_in_archive_order":1,"of":7,"metrics":{"APH/L2":"78.0"},"uses_additional_data":false},{"leaderboard":"/sota/3d-object-detection-on-waymo-pedestrian","task":"3D Object Detection","dataset":"waymo pedestrian","model":"DSVT(val)","rank_in_archive_order":1,"of":7,"metrics":{"APH/L2":"76.4"},"uses_additional_data":false},{"leaderboard":"/sota/3d-object-detection-on-waymo-vehicle","task":"3D Object Detection","dataset":"waymo vehicle","model":"DSVT(val)","rank_in_archive_order":2,"of":8,"metrics":{"APH/L2":"74.1","L1 mAP":"82.1"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/2301.06051","atlas_url":"https://app.syntology.ai/?focus=2301.06051","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}