mlsanityData to 14 Oct 2026

SFMNet: Sparse Focal Modulation for 3D Object Detection

arXiv 2503.12093 ↗First posted 15 Mar 2025cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

We propose SFMNet, a novel 3D sparse detector that combines the efficiency of sparse convolutions with the ability to model long-range dependencies. While traditional sparse convolution techniques efficiently capture local structures, they struggle with modeling long-range relationships. However, capturing long-range dependencies is fundamental for 3D object detection. In contrast, transformers are designed to capture these long-range dependencies through attention mechanisms. But, they come with high computational costs, due to their quadratic query-key-value interactions. Furthermore, directly applying attention to non-empty voxels is inefficient due to the sparse nature of 3D scenes. Our SFMNet is built on a novel Sparse Focal Modulation (SFM) module, which integrates short- and long-range contexts with linear complexity by leveraging a new hierarchical sparse convolution design. This approach enables SFMNet to achieve high detection performance with improved efficiency, making it well-suited for large-scale LiDAR scenes. We show that our detector achieves state-of-the-art performance on autonomous driving datasets.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 63 of its 72 references that are in the map. First posted on arXiv 15 Mar 2025.

Builds on

63 of 72 references.

References · 63

  1. 2017Attention Is All You Needcited by 60,361
  2. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,139
  3. 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,248
  4. 2019nuScenes: A multimodal dataset for autonomous drivingcited by 4,523
  5. 2017PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Spacecited by 4,401
  6. 2016Gaussian Error Linear Units (GELUs)cited by 3,502
  7. 2019Scalability in Perception for Autonomous Driving: Waymo Open Datasetcited by 2,451
  8. 2017Non-local Neural Networkscited by 2,293
  9. 2018PointPillars: Fast Encoders for Object Detection from Point Cloudscited by 1,715
  10. 2020Point Transformercited by 1,470
  11. 2017VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detectioncited by 1,323
  12. 20194D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networkscited by 1,208
38 more
  1. 2020Center-based 3D Object Detection and Trackingcited by 1,144
  2. 2018PointRCNN: 3D Object Proposal Generation and Detection from Point Cloudcited by 941
  3. 2019PV-RCNN: Point-Voxel Feature Set Abstraction for 3D Object Detectioncited by 891
  4. 20173D Semantic Segmentation with Submanifold Sparse Convolutional Networkscited by 748
  5. 2019Argoverse: 3D Tracking and Forecasting with Rich Mapscited by 732
  6. 2023Argoverse 2: Next Generation Datasets for Self-Driving Perception and Forecastingcited by 707
  7. 2017Frustum PointNets for 3D Object Detection from RGB-D Datacited by 640
  8. 2023Point Transformer V3: Simpler, Faster, Strongercited by 589
  9. 2021MetaFormer Is Actually What You Need for Visioncited by 524
  10. 2022TransFusion: Robust LiDAR-Camera Fusion for 3D Object Detection with Transformerscited by 466
  11. 20203DSSD: Point-based 3D Single Stage Object Detectorcited by 440
  12. 2020Voxel R-CNN: Towards High Performance Voxel-based 3D Object Detectioncited by 410
  13. 2019From Points to Parts: 3D Object Detection from Point Cloud with Part-aware and Part-aggregation Networkcited by 368
  14. 2016Target-driven Visual Navigation in Indoor Scenes using Deep Reinforcement Learningcited by 366
  15. 2020Point-GNN: Graph Neural Network for 3D Object Detection in a Point Cloudcited by 335
  16. 2021Voxel Transformer for 3D Object Detectioncited by 247
  17. 2021PV-RCNN++: Point-Voxel Feature Set Abstraction With Local Vector Representation for 3D Object Detectioncited by 232
  18. 2017Submanifold Sparse Convolutional Networkscited by 223
  19. 2023VoxelNeXt: Fully Sparse VoxelNet for 3D Object Detection and Trackingcited by 222
  20. 2021One Million Scenes for Autonomous Driving: ONCE Datasetcited by 218
  21. 20203D Object Detection with Pointformercited by 194
  22. 2021Embracing Single Stride 3D Object Detector with Sparse Transformercited by 193
  23. 2022Focal Modulation Networkscited by 186
  24. 2019End-to-End Multi-View Fusion for 3D Object Detection in LiDAR Point Cloudscited by 171
  25. 2022Scaling Up Your Kernels to 31x31: Revisiting Large Kernel Design in CNNscited by 155
  26. 2022More ConvNets in the 2020s: Scaling up Kernels Beyond 51x51 using Sparsitycited by 142
  27. 2021RangeDet:In Defense of Range View for LiDAR-based 3D Object Detectioncited by 140
  28. 2023DSVT: Dynamic Sparse Voxel Transformer with Rotated Setscited by 122
  29. 2019Fast Point R-CNNcited by 119
  30. 2021SE-SSD: Self-Ensembling Single-Stage Object Detector From Point Cloudcited by 119
  31. 2021RSN: Range Sparse Net for Efficient, Accurate LiDAR 3D Object Detectioncited by 115
  32. 2020CIA-SSD: Confident IoU-Aware Single-Stage Object Detector From Point Cloudcited by 107
  33. 2022CenterFormer: Center-based Transformer for 3D Object Detectioncited by 99
  34. 2022SWFormer: Sparse Window Transformer for 3D Object Detection in Point Cloudscited by 92
  35. 2021AFDetV2: Rethinking the Necessity of the Second Stage for Object Detection from Point Cloudscited by 88
  36. 2022Fully Sparse 3D Object Detectioncited by 84
  37. 2020HVNet: Hybrid Voxel Network for LiDAR Based 3D Object Detectioncited by 82
  38. 2023UniTR: A Unified and Efficient Multi-Modal Transformer for Bird's-Eye-View Representationcited by 74

50 most cited of 63 All 63 (JSON)

Built on by

Not cited yet.