Research map / Detection and segmentation / 3D detection and driving
Spe-BEVHead: Rethinking the Detection Head Design for Bird's-Eye-View Object Detection
Junshu Zhang, Sicheng Zhao, Xin Zhao, Fan Yang, Ruike Chen, Jungong Han, Guiguang Ding
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 dual-branch architecture with auxiliary branch (ablation) | Spe-BEVHead on FastBEV · nuScenes val (end-to-end) | NDS: 34.5 -> 36.6 | Improved |
| add Rotated Box Kernel (ablation) | Spe-BEVHead on FastBEV · nuScenes val (end-to-end) | NDS: 36.6 -> 37.3 | Improved |
| add Local Response Refinement Module (ablation) | Spe-BEVHead on FastBEV · nuScenes val (end-to-end) | NDS: 37.3 -> 37.9 | Improved |
| replace center-based head with Spe-BEVHead | FastBEV · nuScenes val | NDS: 38.4 -> 40.1 | Improved |
| replace center-based head with Spe-BEVHead | FastBEV4D · nuScenes val | NDS: 44.7 -> 44.5 | No clear effect |
| replace center-based head with Spe-BEVHead | BEVStereo4D · nuScenes val | mAP: 38.1 -> 37.6 | Mixed |
| replace center-based head with Spe-BEVHead and drop post-processing | FastBEV · nuScenes val (end-to-end) | mAP: 21.9 -> 26.2 | Improved |
| remove NMS from the center-based head baseline | FastBEV · nuScenes val | mAP: 29.5 -> 21.9 | Hurt |
| replace AMA operator with max pooling in LRRM (ablation) | Spe-BEVHead · nuScenes val (FastBEV) | NDS: 37.9 -> 37.4 | Hurt |
| increase RBK boundary value from 0.1 to 0.3 or 0.5 (ablation) | Rotated Box Kernel · nuScenes val (FastBEV) | NDS: 37.9 -> 37.4 | Hurt |
| stack LRRM three times instead of twice (ablation) | LRRM · nuScenes val (FastBEV) | NDS: 37.9 -> 37.8 | No clear effect |
| enlarge AMA local size to 5x5 (ablation) | LRRM · nuScenes val (FastBEV) | NDS: 37.9 -> 37.3 | Hurt |
Closest papers by meaning
- BEV-MAE: Bird’s Eye View Masked Autoencoders for Point Cloud Pre-training in Autonomous Driving Scenarios (AAAI 2024 · similarity 0.89)
- A Black-Box Evaluation Framework for Semantic Robustness in Bird’s Eye View Detection (AAAI 2025 · similarity 0.89)
- GeoBEV: Learning Geometric BEV Representation for Multi-view 3D Object Detection (AAAI 2025 · similarity 0.88)
- M-BEV: Masked BEV Perception for Robust Autonomous Driving (AAAI 2024 · similarity 0.88)
- UniMODE: Unified Monocular 3D Object Detection (CVPR 2024 · similarity 0.88)
- BEVDiffuser: Plug-and-Play Diffusion Model for BEV Denoising with Ground-Truth Guidance (CVPR 2025 · similarity 0.88)
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.
