mlsanityData to 14 Oct 2026

FSHNet: Fully Sparse Hybrid Network for 3D Object Detection

arXiv 2506.03714 ↗First posted 4 Jun 2025cs.CVcited by 9

Not on the map. Why?

Abstract (arXiv)

Fully sparse 3D detectors have recently gained significant attention due to their efficiency in long-range detection. However, sparse 3D detectors extract features only from non-empty voxels, which impairs long-range interactions and causes the center feature missing. The former weakens the feature extraction capability, while the latter hinders network optimization. To address these challenges, we introduce the Fully Sparse Hybrid Network (FSHNet). FSHNet incorporates a proposed SlotFormer block to enhance the long-range feature extraction capability of existing sparse encoders. The SlotFormer divides sparse voxels using a slot partition approach, which, compared to traditional window partition, provides a larger receptive field. Additionally, we propose a dynamic sparse label assignment strategy to deeply optimize the network by providing more high-quality positive samples. To further enhance performance, we introduce a sparse upsampling module to refine downsampled voxels, preserving fine-grained details crucial for detecting small objects. Extensive experiments on the Waymo, nuScenes, and Argoverse2 benchmarks demonstrate the effectiveness of FSHNet. The code is available at https://github.com/Say2L/FSHNet.

Key facts. Cited by 9 papers in the map (data through 14 Oct 2026). Builds on 49 of its 51 references that are in the map. First posted on arXiv 4 Jun 2025.

Builds on

49 of 51 references.

References · 49

  1. 2017Attention Is All You Needcited by 60,361
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  3. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,139
  4. 2015Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networkscited by 9,639
  5. 2017Mask R-CNNcited by 7,277
  6. 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,248
  7. 2019nuScenes: A multimodal dataset for autonomous drivingcited by 4,523
  8. 2017PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Spacecited by 4,401
  9. 2019Scalability in Perception for Autonomous Driving: Waymo Open Datasetcited by 2,451
  10. 2020Transformers are RNNs: Fast Autoregressive Transformers with Linear Attentioncited by 2,022
  11. 2018PointPillars: Fast Encoders for Object Detection from Point Cloudscited by 1,715
  12. 2020Rethinking Attention with Performerscited by 1,601
37 more
  1. 2019FCOS: Fully Convolutional One-Stage Object Detectioncited by 1,577
  2. 2017VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detectioncited by 1,323
  3. 2019Objects as Pointscited by 1,175
  4. 2020Center-based 3D Object Detection and Trackingcited by 1,144
  5. 2018PointRCNN: 3D Object Proposal Generation and Detection from Point Cloudcited by 941
  6. 2019PV-RCNN: Point-Voxel Feature Set Abstraction for 3D Object Detectioncited by 891
  7. 2023Argoverse 2: Next Generation Datasets for Self-Driving Perception and Forecastingcited by 707
  8. 2019Bridging the Gap Between Anchor-based and Anchor-free Detection via Adaptive Training Sample Selectioncited by 540
  9. 2022TransFusion: Robust LiDAR-Camera Fusion for 3D Object Detection with Transformerscited by 466
  10. 20203DSSD: Point-based 3D Single Stage Object Detectorcited by 440
  11. 2020Voxel R-CNN: Towards High Performance Voxel-based 3D Object Detectioncited by 410
  12. 2019Class-balanced Grouping and Sampling for Point Cloud 3D Object Detectioncited by 350
  13. 2021PV-RCNN++: Point-Voxel Feature Set Abstraction With Local Vector Representation for 3D Object Detectioncited by 232
  14. 2023VoxelNeXt: Fully Sparse VoxelNet for 3D Object Detection and Trackingcited by 222
  15. 2021Embracing Single Stride 3D Object Detector with Sparse Transformercited by 193
  16. 2023EfficientViT: Memory Efficient Vision Transformer with Cascaded Group Attentioncited by 172
  17. 2023FLatten Transformer: Vision Transformer using Focused Linear Attentioncited by 171
  18. 2022Not All Points Are Equal: Learning Highly Efficient Point-based Detectors for 3D LiDAR Point Cloudscited by 131
  19. 2023DSVT: Dynamic Sparse Voxel Transformer with Rotated Setscited by 122
  20. 2021RSN: Range Sparse Net for Efficient, Accurate LiDAR 3D Object Detectioncited by 115
  21. 2021Improving 3D Object Detection with Channel-wise Transformercited by 108
  22. 2022Voxel Set Transformer: A Set-to-Set Approach to 3D Object Detection from Point Cloudscited by 102
  23. 2020AutoAssign: Differentiable Label Assignment for Dense Object Detectioncited by 100
  24. 2022CenterFormer: Center-based Transformer for 3D Object Detectioncited by 99
  25. 2022SWFormer: Sparse Window Transformer for 3D Object Detection in Point Cloudscited by 92
  26. 2021AFDetV2: Rethinking the Necessity of the Second Stage for Object Detection from Point Cloudscited by 88
  27. 2022Fully Sparse 3D Object Detectioncited by 84
  28. 2023FlatFormer: Flattened Window Attention for Efficient Point Cloud Transformercited by 73
  29. 2022LargeKernel3D: Scaling up Kernels in 3D Sparse CNNscited by 72
  30. 2020AFDet: Anchor Free One Stage 3D Object Detectioncited by 63
  31. 2024SAFDNet: A Simple and Effective Network for Fully Sparse 3D Object Detectioncited by 52
  32. 2023HEDNet: A Hierarchical Encoder-Decoder Network for 3D Object Detection in Point Cloudscited by 45
  33. 2024LION: Linear Group RNN for 3D Object Detection in Point Cloudscited by 38
  34. 2023FSD V2: Improving Fully Sparse 3D Object Detection with Virtual Voxelscited by 34
  35. 2023LinK: Linear Kernel for LiDAR-based 3D Perceptioncited by 26
  36. 2024ScatterFormer: Efficient Voxel Transformer with Scattered Linear Attentioncited by 11
  37. 2024DCDet: Dynamic Cross-based 3D Object Detectorcited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 3; 2026: 6.

9 papers.

  1. 2025A Self-Conditioned Representation Guided Diffusion Model for Realistic Text-to-LiDAR Scene Generationcited by 4
  2. 2025Look Before You Fuse: 2D-Guided Cross-Modal Alignment for Robust 3D Detectioncited by 1
  3. 2025Voxel Densification for Serialized 3D Object Detection: Mitigating Sparsity via Pre-serialization Expansioncited by 0
  4. 2026Fore-Mamba3D: Mamba-based Foreground-Enhanced Encoding for 3D Object Detectioncited by 0
  5. 2026Group3D: MLLM-Driven Semantic Grouping for Open-Vocabulary 3D Object Detectioncited by 0
  6. 2026GraphBEV++: Multi-Modal Feature Alignment for Autonomous Drivingcited by 0
  7. 2026T2LDM++: A Self-Conditioned Representation Guided Diffusion Model for Realistic Text-to-LiDAR Scene Generationcited by 0
  8. 2026Horizon3D: Sparse Radar-Camera Fusion for Long-Range 3D Perception in Autonomous Drivingcited by 0
  9. 2026PointLAM: Local Attentive Mamba for Efficient Point-based 3D Object Detectioncited by 0