mlsanityData to 14 Oct 2026

Sparse4D v2: Recurrent Temporal Fusion with Sparse Model

arXiv 2305.14018 ↗First posted 23 May 2023cs.CVcited by 93

Not on the map. Why?

Abstract (arXiv)

Sparse algorithms offer great flexibility for multi-view temporal perception tasks. In this paper, we present an enhanced version of Sparse4D, in which we improve the temporal fusion module by implementing a recursive form of multi-frame feature sampling. By effectively decoupling image features and structured anchor features, Sparse4D enables a highly efficient transformation of temporal features, thereby facilitating temporal fusion solely through the frame-by-frame transmission of sparse features. The recurrent temporal fusion approach provides two main benefits. Firstly, it reduces the computational complexity of temporal fusion from O(T) to O(1), resulting in significant improvements in inference speed and memory usage. Secondly, it enables the fusion of long-term information, leading to more pronounced performance improvements due to temporal fusion. Our proposed approach, Sparse4Dv2, further enhances the performance of the sparse perception algorithm and achieves state-of-the-art results on the nuScenes 3D detection benchmark. Code will be available at urlhttps://github.com/linxuewu/Sparse4D.

Key facts. Cited by 93 papers in the map (data through 14 Oct 2026). Builds on 32 of its 34 references that are in the map. First posted on arXiv 23 May 2023.

Builds on

32 of 34 references.

References · 32

  1. 2017Attention Is All You Needcited by 60,361
  2. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  3. 2020End-to-End Object Detection with Transformerscited by 7,414
  4. 2016Feature Pyramid Networks for Object Detectioncited by 5,036
  5. 2019nuScenes: A multimodal dataset for autonomous drivingcited by 4,523
  6. 2022A ConvNet for the 2020scited by 3,579
  7. 2022BEVFormer: Learning Bird's-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformerscited by 1,218
  8. 2020Lift, Splat, Shoot: Encoding Images From Arbitrary Camera Rigs by Implicitly Unprojecting to 3Dcited by 1,044
  9. 2018MVSNet: Depth Inference for Unstructured Multi-view Stereocited by 788
  10. 2021BEVDet: High-performance Multi-camera 3D Object Detection in Bird-Eye-Viewcited by 713
  11. 2021DETR3D: 3D Object Detection from Multi-view Images via 3D-to-2D Queriescited by 658
  12. 2022BEVDepth: Acquisition of Reliable Depth for Multi-view 3D Object Detectioncited by 572
20 more
  1. 2022PETR: Position Embedding Transformation for Multi-View 3D Object Detectioncited by 552
  2. 2022InternImage: Exploring Large-Scale Vision Foundation Models with Deformable Convolutionscited by 482
  3. 2021FCOS3D: Fully Convolutional One-Stage Monocular 3D Object Detectioncited by 447
  4. 2022PETRv2: A Unified Framework for 3D Perception from Multi-Camera Imagescited by 352
  5. 2019Class-balanced Grouping and Sampling for Point Cloud 3D Object Detectioncited by 350
  6. 2022BEVDet4D: Exploit Temporal Cues in Multi-camera 3D Object Detectioncited by 341
  7. 2023Exploring Object-Centric Temporal Modeling for Efficient Multi-View 3D Object Detectioncited by 315
  8. 2022BEVFormer v2: Adapting Modern Image Backbones to Bird's-Eye-View Recognition via Perspective Supervisioncited by 302
  9. 2021Is Pseudo-Lidar needed for Monocular 3D Object detection?cited by 249
  10. 2020SMOKE: Single-Stage Monocular 3D Object Detection via Keypoint Estimationcited by 204
  11. 2022Time Will Tell: New Outlooks and A Baseline for Temporal Multi-View 3D Object Detectioncited by 178
  12. 2019An Energy and GPU-Computation Efficient Backbone Network for Real-Time Object Detectioncited by 163
  13. 2022Sparse4D: Multi-view 3D Object Detection with Sparse Spatial-Temporal Fusioncited by 162
  14. 2019Monocular 3D Object Detection with Pseudo-LiDAR Point Cloudcited by 113
  15. 2020Fast-MVSNet: Sparse-to-Dense Multi-View Stereo With Learned Propagation and Gauss-Newton Refinementcited by 105
  16. 2022BEVStereo: Enhancing Depth Estimation in Multi-view 3D Object Detection with Dynamic Temporal Stereocited by 90
  17. 2022BEVPoolv2: A Cutting-edge Implementation of BEVDet Toward Deploymentcited by 79
  18. 2022BEVDistill: Cross-Modal BEV Distillation for Multi-View 3D Object Detectioncited by 66
  19. 2023Exploring Recurrent Long-term Temporal Fusion for Multi-view 3D Perceptioncited by 65
  20. 2023Temporal Enhanced Training of Multi-view 3D Object Detector via Historical Object Predictioncited by 49

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 9; 2024: 36; 2025: 28; 2026: 20.

93 papers.

Sort:

By area:

  1. 2023Is Ego Status All You Need for Open-Loop End-to-End Autonomous Driving?cited by 242
  2. 2024SparseDrive: End-to-End Autonomous Driving via Sparse Scene Representationcited by 241
  3. 2023SparseBEV: High-Performance Sparse 3D Object Detection from Multi-Camera Videoscited by 150
  4. 2023StreamMapNet: Streaming Mapping Network for Vectorized Online HD Map Constructioncited by 145
  5. 2023FB-BEV: BEV Representation from Forward-Backward View Transformationscited by 98
  6. 2023Far3D: Expanding the Horizon for Surround-view 3D Object Detectioncited by 70
  7. 2023Sparse4D v3: Advancing End-to-End 3D Detection and Trackingcited by 70
  8. 2024MapTracker: Tracking with Strided Memory Fusion for Consistent Vector HD Mappingcited by 68
  9. 2024OPUS: Occupancy Prediction Using a Sparse Setcited by 56
  10. 2024SparseAD: Sparse Query-Centric Paradigm for Efficient End-to-End Autonomous Drivingcited by 47
  11. 2024Stream Query Denoising for Vectorized HD Map Constructioncited by 43
  12. 2024Ray Denoising: Depth-aware Hard Negative Sampling for Multi-view 3D Object Detectioncited by 36
38 more
  1. 2023BEVNeXt: Reviving Dense BEV Frameworks for 3D Object Detectioncited by 35
  2. 2025HiP-AD: Hierarchical and Multi-Granularity Planning with Deformable Attention for Autonomous Driving in a Single Decodercited by 35
  3. 2024ADA-Track++: End-to-End Multi-Camera 3D Multi-Object Tracking with Alternating Detection and Associationcited by 33
  4. 2024SparseLIF: High-Performance Sparse LiDAR-Camera Fusion for 3D Object Detectioncited by 25
  5. 2024MV2DFusion: Leveraging Modality-Specific Object Semantics for Multi-Modal 3D Detectioncited by 23
  6. 2024PreSight: Enhancing Autonomous Vehicle Perception with City-Scale NeRF Priorscited by 21
  7. 2024Is a 3D-Tokenized LLM the Key to Reliable Autonomous Driving?cited by 20
  8. 2024Neural Radiance Field in Autonomous Driving: A Surveycited by 19
  9. 2024HENet: Hybrid Encoding for End-to-end Multi-task 3D Perception from Multi-view Camerascited by 16
  10. 2024UAV3D: A Large-scale 3D Perception Benchmark for Unmanned Aerial Vehiclescited by 16
  11. 2024CLIP-BEVFormer: Enhancing Multi-View Image-Based BEV Detector with Ground Truth Flowcited by 14
  12. 2025CoCMT: Communication-Efficient Cross-Modal Transformer for Collaborative Perceptioncited by 13
  13. 2026DVGT-2: Vision-Geometry-Action Model for Autonomous Driving at Scalecited by 13
  14. 2024OPEN: Object-wise Position Embedding for Multi-view 3D Object Detectioncited by 12
  15. 2024PrevPredMap: Exploring Temporal Modeling with Previous Predictions for Online Vectorized HD Map Constructioncited by 12
  16. 2024GeoBEV: Learning Geometric BEV Representation for Multi-view 3D Object Detectioncited by 11
  17. 2024SimPB: A Single Model for 2D and 3D Object Detection from Multiple Camerascited by 9
  18. 2024RayFormer: Improving Query-Based Multi-Camera 3D Object Detection via Ray-Centric Strategiescited by 9
  19. 2024Make Your ViT-based Multi-view 3D Detectors Faster via Token Compressioncited by 9
  20. 2025Rethinking Temporal Fusion with a Unified Gradient Descent View for 3D Semantic Occupancy Predictioncited by 9
  21. 2025Bridging Perspectives: Foundation Model Guided BEV Maps for 3D Object Detection and Trackingcited by 9
  22. 2026VGGT-Det: Mining VGGT Internal Priors for Sensor-Geometry-Free Multi-View Indoor 3D Object Detectioncited by 9
  23. 2024MIM4D: Masked Modeling with Multi-View Video for Autonomous Driving Representation Learningcited by 8
  24. 2025HisTrackMap: Global Vectorized High-Definition Map Construction via History Map Trackingcited by 8
  25. 2025InteractionMap: Improving Online Vectorized HDMap Construction with Interactioncited by 8
  26. 2024Scaling Multi-Camera 3D Object Detection through Weak-to-Strong Elicitingcited by 7
  27. 2025HENet++: Hybrid Encoding and Multi-task Learning for 3D Perception and End-to-end Autonomous Drivingcited by 6
  28. 2023QE-BEV: Query Evolution for Bird's Eye View Object Detection in Varied Contextscited by 5
  29. 2024RopeBEV: A Multi-Camera Roadside Perception Network in Bird's-Eye-Viewcited by 5
  30. 2025MDP: Multidimensional Vision Model Pruning with Latency Constraintcited by 5
  31. 2025Trends in Motion Prediction Toward Deployable and Generalizable Autonomy: A Revisit and Perspectivescited by 5
  32. 2025A Survey of Multi-sensor Fusion Perception for Embodied AI: Background, Methods, Challenges and Prospectscited by 5
  33. 2025ForeSight: Multi-View Streaming Joint Object Detection and Trajectory Forecastingcited by 5
  34. 2025Spatial Retrieval Augmented Autonomous Drivingcited by 5
  35. 2024Multi-Dimensional Pruning: Joint Channel, Layer and Block Pruning with Latency Constraintcited by 4
  36. 2024Learning High-resolution Vector Representation from Multi-Camera Images for 3D Object Detectioncited by 4
  37. 2024GenMapping: Unleashing the Potential of Inverse Perspective Mapping for Robust Online HD Map Constructioncited by 4
  38. 2025TQD-Track: Temporal Query Denoising for 3D Multi-Object Trackingcited by 4

50 most cited of 93 All 93 (JSON)