Research map / 3D vision / 3D generation
DualBEV: Unifying Dual View Transformation with Probabilistic Correspondences
Peidong Li, Wancheng Shen, Qihao Huang, Dixiao Cui
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 ProbNet BEV probability (ablation) | BEVDepth baseline · nuScenes val | NDS: 40.7 -> 41.1 | Improved |
| add HeightTrans (3D-to-2D features) (ablation) | Prob-LSS · nuScenes val | NDS: 41.1 -> 41.5 | Improved |
| add CAF module for soft feature selection (ablation) | DualBEV · nuScenes val | NDS: 41.5 -> 41.6 | Improved |
| add SAE module (spatial attention for BEV probability) (ablation) | DualBEV · nuScenes val | NDS: 41.6 -> 42.5 | Improved |
| add SAE module on mAP (ablation) | DualBEV · nuScenes val | mAP: 35.3 -> 35.2 | No clear effect |
| use projection probability P_proj (disable by uniform depth) (ablation) | DualBEV · nuScenes val | NDS: 40.0 -> 50.4 | Improved |
| add both image and BEV probabilities to projection probability (ablation) | DualBEV · nuScenes val | NDS: 48.8 -> 50.4 | Improved |
| replace BEV Pooling with Prob-Sampling | BEVDet · nuScenes val | NDS: 38.2 -> 39.0 | Improved |
| replace BEV Pooling with Prob-Sampling on mAP | BEVDet · nuScenes val | mAP: 30.7 -> 30.6 | No clear effect |
| use multi-resolution height sampling (13 points) instead of uniform sampling (ablation) | HeightTrans · nuScenes val | NDS: 41.9 -> 42.5 (vs 4 uniform), 42.3 -> 42.5 (vs 16 uniform) | Improved |
| use DFF one-stage fusion instead of FB-BEV style refinement (ablation) | DualBEV fusion · nuScenes val | NDS: 41.7 -> 42.5 | Improved |
| extend probability filtering from image space (SA-BEV) to BEV space (Prob-LSS) (ablation) | LSS · nuScenes val | mAP: 30.9 -> 31.5 | Improved |
Closest papers by meaning
- Delving Into the Devils of Bird’s-Eye-View Perception: A Review, Evaluation and Recipe (TPAMI 2023 · similarity 0.88)
- Toward Real-world BEV Perception: Depth Uncertainty Estimation via Gaussian Splatting (CVPR 2025 · similarity 0.88)
- FB-BEV: BEV Representation from Forward-Backward View Transformations (ICCV 2023 · similarity 0.88)
- A Black-Box Evaluation Framework for Semantic Robustness in Bird’s Eye View Detection (AAAI 2025 · similarity 0.86)
- Improving Bird's Eye View Semantic Segmentation by Task Decomposition (CVPR 2024 · similarity 0.85)
- MamBEV: Enabling State Space Models to Learn Birds-Eye-View Representations (ICLR 2025 · similarity 0.84)
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.
