Research map / Detection and segmentation / 3D detection and driving
Learning High-resolution Vector Representation from Multi-Camera Images for 3D Object Detection
Zhili Chen, Shuangjie Xu, Maosheng Ye, Zian Qian, Xiaoyi Zou, Dit-Yan Yeung, Qifeng Chen
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 vector query representation to BEVFormer-S | BEVFormer-S · nuScenes val | NDS: 47.9 -> 51.0 | Improved |
| add vector representation to DFA3D-S | DFA3D-S · nuScenes val | NDS: 50.1 -> 51.4 | Improved |
| add vector representation to BEVFormer-B | BEVFormer-B · nuScenes val | NDS: 51.7 -> 53.2 | Improved |
| use vector queries with coarser BEV grid instead of larger BEV grid | BEVFormer · nuScenes val | Memory: 5.58 -> 4.83 | Improved |
| increase BEV resolution in plain BEVFormer (ablation) | BEVFormer · nuScenes val | NDS: 51.7 -> 51.9 | No clear effect |
| increase vector dimension from 200 to 450 at fixed 200x200 BEV (ablation) | VectorFormer · nuScenes val | NDS: 52.8 -> 53.2 | Improved |
| add heatmap supervision for BEV features (ablation) | VectorFormer-S · nuScenes val (12 epochs) | NDS: 48.4 -> 48.9 | Improved |
| add vector representation for spatial modeling (ablation) | VectorFormer-S · nuScenes val (12 epochs) | NDS: 48.9 -> 49.5 | Improved |
| add temporal modeling of vector queries (ablation) | VectorFormer-S · nuScenes val (12 epochs) | NDS: 49.5 -> 49.7 | Improved |
| combine x/y vectors by addition instead of multiplication (ablation) | VectorFormer-S · nuScenes val (12 epochs) | NDS: 49.4 -> 49.7 | Improved |
| increase top-k proposals from 2 to 4 (ablation) | VectorFormer-S · nuScenes val (12 epochs) | NDS: 49.5 -> 50.0 | Improved |
| add deformable offset prediction (ablation) | VectorFormer-S · nuScenes val (12 epochs) | mAP: 39.2 -> 39.8 | Improved |
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.
- BEVFormer — BEVFormer: Learning Bird’s-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformers (ECCV 2022) · better on 2 of 2 shared results
- PETRv2 — PETR: Position Embedding Transformation for Multi-View 3D Object Detection (ECCV 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
- GeoBEV: Learning Geometric BEV Representation for Multi-view 3D Object Detection (AAAI 2025 · similarity 0.90)
- OA-DET3D: Embedding Object Awareness As A General Plug-in for Multi-Camera 3D Object Detection (IJCV 2025 · similarity 0.88)
- BEVNeXt: Reviving Dense BEV Frameworks for 3D Object Detection (CVPR 2024 · similarity 0.88)
- SparseBEV: A Fully Sparse Framework for Multi-View 3D Object Detection (TPAMI 2026 · similarity 0.88)
- RaCFormer: Towards High-Quality 3D Object Detection via Query-based Radar-Camera Fusion (CVPR 2025 · similarity 0.88)
- Towards Generalizable Multi-Camera 3D Object Detection via Perspective Rendering (AAAI 2025 · 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.
