Research map / Detection and segmentation / 3D detection and driving
BEVFormer: Learning Bird’s-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformers
Zhiqi Li, Wenhai Wang, Hongyang Li, Enze Xie, Chonghao Sima, Tong Lu, Yu Qiao, Jifeng Dai
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 temporal self-attention (ablation) | BEVFormer-S · nuScenes val | NDS: 0.448 -> 0.517 | Improved |
| add temporal self-attention (ablation) | BEVFormer-S · nuScenes val | mAVE: 0.802 -> 0.394 | Improved |
| add temporal self-attention (ablation) | BEVFormer-S R101 · nuScenes test | NDS: 0.462 -> 0.535 | Improved |
| use local deformable spatial attention instead of global attention (ablation) | BEVFormer-S · nuScenes val | NDS: 0.404 -> 0.448 | Improved |
| use local deformable attention instead of reference-point-only interaction (ablation) | BEVFormer-S · nuScenes val | NDS: 0.423 -> 0.448 | Improved |
| increase training frames from 1 to 4 (ablation) | BEVFormer · nuScenes val | NDS: 0.448 -> 0.517 | Improved |
| increase training frames from 4 to 5 (ablation) | BEVFormer · nuScenes val | NDS: 0.517 -> 0.517 | No clear effect |
| add ego-motion alignment of history BEV features (ablation) | BEVFormer · nuScenes val | NDS: 51.0% NDS -> 51.7% NDS | Improved |
| add random sampling of 4 frames from 5 (ablation) | BEVFormer · nuScenes val | NDS: 51.3% NDS -> 51.7% NDS | Improved |
| use both BEV queries and history BEV features to predict offsets (ablation) | BEVFormer · nuScenes val | NDS: 51.3% NDS -> 51.7% NDS | Improved |
| reduce encoder to 1 layer (config C) (ablation) | BEVFormer · nuScenes val | NDS: 0.517 -> 0.501 | Hurt |
| single-scale features, 100x100 BEV, 1 layer (config D) (ablation) | BEVFormer · nuScenes val | NDS: 0.517 -> 0.478 | Hurt |
Later papers that compared with this method
32 later accepted papers report BEVFormer as a baseline in their results tables; 30 of them report a better number on most of the results they share with it.
- SemLT3D: Semantic-Guided Expert Distillation for Camera-only Long-Tailed 3D Object Detection (CVPR 2026 · better on 4 of 4 shared results)
- CycleBEV: Regularizing View Transformation Networks via View Cycle Consistency for Bird's-Eye-View Semantic Segmentation (CVPR 2026 · better on 3 of 3 shared results)
- MTA: Multimodal Task Alignment for BEV Perception and Captioning (CVPR 2026 · better on 2 of 2 shared results)
- RayD3D: Distilling Depth Knowledge Along the Ray for Robust Multi-View 3D Object Detection (AAAI 2026 · better on 2 of 2 shared results)
- GUIDE: Gaussian Unified Instance Detection for Enhanced Obstacle Perception in Autonomous Driving (AAAI 2026 · better on 0 of 2 shared results)
- CLIPDet3D: Vision-Language Collaborative Distillation for 3D Object Detection (AAAI 2026 · better on 2 of 2 shared results)
- BEVDiffuser: Plug-and-Play Diffusion Model for BEV Denoising with Ground-Truth Guidance (CVPR 2025 · better on 8 of 8 shared results)
- CorrBEV: Multi-View 3D Object Detection by Correlation Learning with Multi-modal Prototypes (CVPR 2025 · better on 4 of 4 shared results)
24 more not listed.
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
- Improving Bird's Eye View Semantic Segmentation by Task Decomposition (CVPR 2024 · similarity 0.91)
- M-BEV: Masked BEV Perception for Robust Autonomous Driving (AAAI 2024 · similarity 0.90)
- CLIP-BEVFormer: Enhancing Multi-View Image-Based BEV Detector with Ground Truth Flow (CVPR 2024 · similarity 0.89)
- MamBEV: Enabling State Space Models to Learn Birds-Eye-View Representations (ICLR 2025 · similarity 0.89)
- BEV-MAE: Bird’s Eye View Masked Autoencoders for Point Cloud Pre-training in Autonomous Driving Scenarios (AAAI 2024 · similarity 0.89)
- Map It Anywhere: Empowering BEV Map Prediction using Large-scale Public Datasets (NeurIPS 2024 · 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.
