DuoSpaceNet: Leveraging Both Bird's-Eye-View and Perspective View Representations for 3D Object Detection
Not on the map. Why?
Abstract (arXiv)
Multi-view camera-only 3D object detection largely follows two primary paradigms: exploiting bird's-eye-view (BEV) representations or focusing on perspective-view (PV) features, each with distinct advantages. Although several recent approaches explore combining BEV and PV, many rely on partial fusion or maintain separate detection heads. In this paper, we propose DuoSpaceNet, a novel framework that fully unifies BEV and PV feature spaces within a single detection pipeline for comprehensive 3D perception. Our design includes a decoder to integrate BEV and PV features into unified detection queries, as well as a feature enhancement strategy that enriches different feature representations. In addition, DuoSpaceNet can be extended to handle multi-frame inputs, enabling more robust temporal analysis. Extensive experiments on nuScenes dataset show that DuoSpaceNet surpasses both BEV-based baselines (e.g., BEVFormer) and PV-based baselines (e.g., Sparse4D) in 3D object detection and BEV map segmentation, verifying the effectiveness of our proposed design.
Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 59 of its 67 references that are in the map. First posted on arXiv 17 May 2024.
Builds on
59 of 67 references.
References · 59
- 2017Attention Is All You Needcited by 60,361
- 2015Deep Residual Learning for Image Recognitioncited by 52,096
- 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,713
- 2020End-to-End Object Detection with Transformerscited by 7,414
- 2016SGDR: Stochastic Gradient Descent with Warm Restartscited by 5,045
- 2016Feature Pyramid Networks for Object Detectioncited by 5,036
- 2019nuScenes: A multimodal dataset for autonomous drivingcited by 4,523
- 2020Deformable DETR: Deformable Transformers for End-to-End Object Detectioncited by 3,422
- 2016V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentationcited by 2,292
- 2022BEVFormer: Learning Bird's-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformerscited by 1,218
- 2020Center-based 3D Object Detection and Trackingcited by 1,144
- 2020Lift, Splat, Shoot: Encoding Images From Arbitrary Camera Rigs by Implicitly Unprojecting to 3Dcited by 1,044
38 more
- 2021BEVDet: High-performance Multi-camera 3D Object Detection in Bird-Eye-Viewcited by 713
- 2023Argoverse 2: Next Generation Datasets for Self-Driving Perception and Forecastingcited by 707
- 2018Deformable ConvNets v2: More Deformable, Better Resultscited by 670
- 2021DETR3D: 3D Object Detection from Multi-view Images via 3D-to-2D Queriescited by 658
- 2017Generalised Dice overlap as a deep learning loss function for highly unbalanced segmentationscited by 620
- 2022BEVDepth: Acquisition of Reliable Depth for Multi-view 3D Object Detectioncited by 572
- 2022PETR: Position Embedding Transformation for Multi-View 3D Object Detectioncited by 552
- 2022PETRv2: A Unified Framework for 3D Perception from Multi-Camera Imagescited by 352
- 2022BEVDet4D: Exploit Temporal Cues in Multi-camera 3D Object Detectioncited by 341
- 2023Exploring Object-Centric Temporal Modeling for Efficient Multi-View 3D Object Detectioncited by 315
- 2022BEVFormer v2: Adapting Modern Image Backbones to Bird's-Eye-View Recognition via Perspective Supervisioncited by 302
- 2022Cross-view Transformers for real-time Map-view Semantic Segmentationcited by 264
- 2023VoxelNeXt: Fully Sparse VoxelNet for 3D Object Detection and Trackingcited by 222
- 2022Unifying Voxel-based Representation with Transformer for 3D Object Detectioncited by 202
- 2019CenterMask : Real-Time Anchor-Free Instance Segmentationcited by 201
- 2022Time Will Tell: New Outlooks and A Baseline for Temporal Multi-View 3D Object Detectioncited by 178
- 2019Cross-view Semantic Segmentation for Sensing Surroundingscited by 174
- 2019End-to-End Multi-View Fusion for 3D Object Detection in LiDAR Point Cloudscited by 171
- 2022Sparse4D: Multi-view 3D Object Detection with Sparse Spatial-Temporal Fusioncited by 162
- 2022M2BEV: Multi-Camera Joint 3D Detection and Segmentation with Unified Birds-Eye View Representationcited by 156
- 2023SparseBEV: High-Performance Sparse 3D Object Detection from Multi-Camera Videoscited by 150
- 2022PolarFormer: Multi-camera 3D Object Detection with Polar Transformercited by 147
- 2022Simple-BEV: What Really Matters for Multi-Sensor BEV Perception?cited by 128
- 2022BEVSegFormer: Bird's Eye View Semantic Segmentation From Arbitrary Camera Rigscited by 123
- 2023Sparse4D v2: Recurrent Temporal Fusion with Sparse Modelcited by 93
- 2018Monocular Semantic Occupancy Grid Mapping with Convolutional Variational Encoder-Decoder Networkscited by 84
- 2022LidarMultiNet: Towards a Unified Multi-Task Network for LiDAR Perceptioncited by 74
- 2015Sparse 3D convolutional neural networkscited by 72
- 2023Sparse4D v3: Advancing End-to-End 3D Detection and Trackingcited by 70
- 2023Exploring Recurrent Long-term Temporal Fusion for Multi-view 3D Perceptioncited by 65
- 2022Polar Parametrization for Vision-based Surround-View 3D Detectioncited by 53
- 2023Temporal Enhanced Training of Multi-view 3D Object Detector via Historical Object Predictioncited by 49
- 2023CAPE: Camera View Position Embedding for Multi-View 3D Object Detectioncited by 40
- 2023Object as Query: Lifting any 2D Object Detector to 3D Detectioncited by 38
- 2022TiG-BEV: Multi-view BEV 3D Object Detection via Target Inner-Geometry Learningcited by 37
- 2022Addressing Optimism Bias in Sequence Modeling for Reinforcement Learningcited by 22
- 2022A Simple Baseline for Multi-Camera 3D Object Detectioncited by 16
- 2020MEBOW: Monocular Estimation of Body Orientation In the Wildcited by 14
50 most cited of 59 All 59 (JSON)
Built on by
Citing papers per year: 2024: 1; 2025: 1.
2 papers.