Research map / Detection and segmentation / 3D detection and driving
GraphBEV: Towards Robust BEV Feature Alignment for Multi-Modal 3D Object Detection
Ziying Song, Lei Yang, Shaoqing Xu, Lin Liu, Dongyang Xu, Caiyan Jia, Feiyang Jia, Li Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add LocalAlign (neighbor depth via KD-tree graph) to BEVFusion (ablation) | BEVFusion · nuScenes val (noisy misalignment) | mAP: 60.8 -> 67.0 | Improved |
| add GlobalAlign (learnable offset alignment) to BEVFusion (ablation) | BEVFusion · nuScenes val (noisy misalignment) | mAP: 60.8 -> 63.1 | Improved |
| add both LocalAlign and GlobalAlign (ablation) | BEVFusion · nuScenes val (noisy misalignment) | mAP: 60.8 -> 69.1 | Improved |
| add LocalAlign (ablation) | BEVFusion · nuScenes val (clean) | mAP: 68.5 -> 69.7 | Improved |
| add GlobalAlign (ablation) | BEVFusion · nuScenes val (clean) | mAP: 68.5 -> 68.9 | Improved |
| add both modules, latency cost (ablation) | BEVFusion · nuScenes val (clean) | latency (ms): 133.2 -> 140.9 | Hurt |
| vary number of graph neighbors K_graph from 5 to 25 (ablation) | LocalAlign · nuScenes val (noisy misalignment) | mAP: 67.1 (K=5), 69.1 (K=8), 69.8 (K=12), 68.8 (K=16), 67.1 (K=25) | Mixed |
| apply GraphBEV to BEV map segmentation | BEVFusion · nuScenes val BEV segmentation | mean IoU: 62.7 -> 63.3 | Improved |
| GraphBEV for night scenes | BEVFusion · nuScenes val night | mAP: 42.8 -> 45.1 | Improved |
| GraphBEV on near objects | BEVFusion · nuScenes val Near (0-20m) | mAP: 79.4 -> 78.6 | Hurt |
Later papers that compared with this method
2 later accepted papers report GraphBEV as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Look Before You Fuse: 2D-Guided Cross-Modal Alignment for Robust 3D Detection (CVPR 2026 · better on 4 of 4 shared results)
- DeGuNet: Depth-Guided Ultra-Compact Backbones for Efficient LiDAR-Camera 3D Detection (ECCV 2026 · better on 1 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- TransFusion — TransFusion: Robust LiDAR-Camera Fusion for 3D Object Detection With Transformers (CVPR 2022) · better on 2 of 2 shared results
- DeepInteraction — DeepInteraction: 3D Object Detection via Modality Interaction (NeurIPS 2022) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- GraphAlign: Enhancing Accurate Feature Alignment by Graph matching for Multi-Modal 3D Object Detection (ICCV 2023 · similarity 0.91)
- GeoBEV: Learning Geometric BEV Representation for Multi-view 3D Object Detection (AAAI 2025 · similarity 0.90)
- UniMODE: Unified Monocular 3D Object Detection (CVPR 2024 · similarity 0.88)
- RobuRCDet: Enhancing Robustness of Radar-Camera Fusion in Bird's Eye View for 3D Object Detection (ICLR 2025 · similarity 0.87)
- Towards Generalizable Multi-Camera 3D Object Detection via Perspective Rendering (AAAI 2025 · similarity 0.87)
- RCBEVDet: Radar-camera Fusion in Bird's Eye View for 3D Object Detection (CVPR 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
