Research map / Detection and segmentation / 3D detection and driving
3D-CVF: Generating Joint Camera and LiDAR Features Using Cross-View Spatial Feature Fusion for 3D Object Detection
Jin Hyeok Yoo, Yecheol Kim, Jisong Kim, Jun Won Choi
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 naive camera features to LiDAR features (ablation) | LiDAR-only baseline · KITTI valid (Car) | 3D AP (Easy/Mod./Hard): 88.35 -> 88.74; 78.31 -> 78.54; 77.08 -> 77.25 | No clear effect |
| add adaptive gated fusion network (ablation) | naive fusion · KITTI valid (Car) | 3D AP (Easy/Mod./Hard): +0.54%, +0.87%, +0.79% | Improved |
| add cross-view feature mapping (ablation) | 3D-CVF with gated fusion · KITTI valid (Car) | 3D AP (Easy/Mod./Hard): +0.5%, +0.06%, +0.15% | Improved |
| add 3D RoI fusion-based refinement (ablation) | 3D-CVF with gated fusion and cross-view mapping · KITTI valid (Car) | 3D AP (Easy/Mod./Hard): +0.28%, +0.63%, +0.45% | Improved |
| add the full camera fusion strategy (ablation) | LiDAR-only baseline · KITTI valid (Car) | 3D AP (Easy/Mod./Hard): +1.32%, +1.57%, +1.39% | Improved |
| add camera fusion | LiDAR-only baseline · nuScenes val | mAP: 39.43 -> 42.17 | Improved |
| add camera fusion | LiDAR-only baseline · nuScenes val | NDS: 46.21 -> 49.78 | Improved |
| add camera fusion, objects 0-20m | LiDAR-only baseline · KITTI valid (0-20m) | 3D AP: 89.86 -> 90.02 (+0.16) | Improved |
| add camera fusion, objects 20-40m | LiDAR-only baseline · KITTI valid (20-40m) | 3D AP: 76.72 -> 79.73 (+3.01) | Improved |
| add camera fusion, objects 40-70m | LiDAR-only baseline · KITTI valid (40-70m) | 3D AP: 30.57 -> 35.86 (+5.29) | Improved |
| compare 3D-CVF with LiDAR-based STD at hard difficulty | 3D-CVF · KITTI test (Car) | 3D AP Hard: 73.11 vs 75.09 | Hurt |
| compare 3D-CVF with best fusion method UberATG-MMF at hard difficulty | 3D-CVF · KITTI test (Car) | 3D AP Hard: 70.22 -> 73.11 | Improved |
Later papers that compared with this method
1 later accepted paper reports 3D-CVF as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Deformable Feature Aggregation for Dynamic Multi-modal 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
- VISTA: Boosting 3D Object Detection via Dual Cross-VIew SpaTial Attention (CVPR 2022 · similarity 0.91)
- ObjectFusion: Multi-modal 3D Object Detection with Object-Centric Fusion (ICCV 2023 · similarity 0.91)
- CCF: Complementary Collaborative Fusion for Domain Generalized Multi-Modal 3D Object Detection (CVPR 2026 · similarity 0.91)
- LoGoNet: Towards Accurate 3D Object Detection With Local-to-Global Cross-Modal Fusion (CVPR 2023 · similarity 0.90)
- MGAF: LiDAR-Camera 3D Object Detection With Multiple Guidance and Adaptive Fusion (TPAMI 2025 · similarity 0.90)
- SimDistill: Simulated Multi-Modal Distillation for BEV 3D Object Detection (AAAI 2024 · similarity 0.89)
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.
