Research map / Detection and segmentation / 3D detection and driving
CluB: Cluster Meets BEV for LiDAR-Based 3D Object Detection
Yingjie Wang, Jiajun Deng, Yuenan Hou, Yao Li, Yu Zhang, Jianmin Ji, Wanli Ouyang, Yanyong Zhang
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 CFD module (soft cluster-to-BEV diffusion) (ablation) | BEV baseline · WOD val 20% | mAPH/L2: 60.5 -> 61.9 | Improved |
| add imitation loss on top of CFD (ablation) | CluB · WOD val 20% | mAPH/L2: 61.9 -> 62.5 | Improved |
| add CQG module (ablation) | BEV baseline · WOD val 20% | mAPH/L2: 60.5 -> 61.0 | Improved |
| add direction loss to CQG (ablation) | CluB · WOD val 20% | mAPH/L2: 61.0 -> 61.6 | Improved |
| combine all four components (ablation) | BEV baseline · WOD val 20% | mAPH/L2: 60.5 -> 63.9 | Improved |
| replace direct concatenation with 2D convolution in CFD (ablation) | CFD module · WOD val Vehicle | APH/L2: 61.9 -> 61.4 | Hurt |
| replace concatenation with class-agnostic max pooling (ablation) | CFD module · WOD val Vehicle vs Pedestrian | APH/L2: 61.9 -> 61.0 (Veh), 63.0 -> 63.4 (Ped) | Mixed |
| make diffusion class-aware (ablation) | CFD max pooling · WOD val Vehicle | APH/L2: 61.0 -> 62.9 | Improved |
| make diffusion class-aware (ablation) | CFD max pooling · WOD val Pedestrian | APH/L2: 63.4 -> 64.4 | Improved |
| add cluster branch with KNN-4 aggregation (ablation) | BEV detector · WOD val Pedestrian | APH/L2: 60.4 -> 64.3 | Improved |
| use KNN-8 instead of KNN-4 aggregation (ablation) | cluster branch · WOD val Pedestrian | APH/L2: 64.3 -> 59.9 | Hurt |
| use KNN-8 instead of KNN-4 aggregation (ablation) | cluster branch · WOD val Vehicle | APH/L2: 61.4 -> 63.2 | 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.
- ConQueR* — ConQueR: Query Contrast Voxel-DETR for 3D Object Detection (CVPR 2023) · better on 5 of 6 shared results
- VoxelNext — VoxelNeXt: Fully Sparse VoxelNet for 3D Object Detection and Tracking (CVPR 2023) · better on 3 of 3 shared results
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
- ObjectFusion: Multi-modal 3D Object Detection with Object-Centric Fusion (ICCV 2023 · similarity 0.90)
- BEVDistill: Cross-Modal BEV Distillation for Multi-View 3D Object Detection (ICLR 2023 · similarity 0.90)
- GeoBEV: Learning Geometric BEV Representation for Multi-view 3D Object Detection (AAAI 2025 · similarity 0.90)
- BEV-MAE: Bird’s Eye View Masked Autoencoders for Point Cloud Pre-training in Autonomous Driving Scenarios (AAAI 2024 · similarity 0.89)
- SAFDNet: A Simple and Effective Network for Fully Sparse 3D Object Detection (CVPR 2024 · similarity 0.89)
- SimDistill: Simulated Multi-Modal Distillation for BEV 3D Object Detection (AAAI 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.
