Research map / Detection and segmentation / 3D detection and driving
Every View Counts: Cross-View Consistency in 3D Object Detection with Hybrid-Cylindrical-Spherical Voxelization
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 RV detection loss as auxiliary supervision without consistency (ablation) | CVCNet BEV branch · NuScenes val | mAP: 51.0 -> 49.8 | Hurt |
| add cross-view consistency losses (both directions) (ablation) | CVCNet BEV inference · NuScenes val | mAP: 51.0 -> 53.9 | Helped |
| add RV->BEV consistency loss (ablation) | CVCNet BEV inference · NuScenes val | mAP: 49.8 -> 53.4 | Helped |
| add BEV->RV consistency loss (ablation) | CVCNet BEV inference · NuScenes val | mAP: 49.8 -> 52.6 | Helped |
| add consistency losses (ablation) | CVCNet RV inference · NuScenes val | mAP: 47.7 -> 48.0 | Helped |
| blend BEV and RV->BEV scores at inference (ablation) | CVCNet · NuScenes val | mAP: 53.9 -> 54.6 | Helped |
| replace learnable transformer with average pooling (ablation) | CVCNet cross-view transformer · NuScenes val | mAP: drops by 44% | Hurt |
| replace learnable transformer with max pooling (ablation) | CVCNet cross-view transformer · NuScenes val | mAP: drops by 4% | Hurt |
| use separate backbones instead of shared (ablation) | CVCNet · NuScenes val | mAP: 54.2 vs 54.6 | No clear effect |
| use shared backbone instead of separate (ablation) | CVCNet · NuScenes val | FPS: 3 -> 11 | Helped |
Closest papers by meaning
- VISTA: Boosting 3D Object Detection via Dual Cross-VIew SpaTial Attention (CVPR 2022 · similarity 0.89)
- BEVSpread: Spread Voxel Pooling for Bird's-Eye-View Representation in Vision-based Roadside 3D Object Detection (CVPR 2024 · similarity 0.88)
- Radial Scaling Voxelization for Accurate Small Object 3D Detection (ICML 2026 · similarity 0.88)
- Voxel or Pillar: Exploring Efficient Point Cloud Representation for 3D Object Detection (AAAI 2024 · similarity 0.87)
- GeoBEV: Learning Geometric BEV Representation for Multi-view 3D Object Detection (AAAI 2025 · similarity 0.87)
- Towards Generalizable Multi-Camera 3D Object Detection via Perspective Rendering (AAAI 2025 · similarity 0.86)
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.