Research map / Detection and segmentation / 3D detection and driving
Boosting Multi-View Indoor 3D Object Detection via Adaptive 3D Volume Construction
Runmin Zhang, Zhu Yu, Si-Yuan Cao, Lingyu Zhu, Guangyi Zhang, Xiaokai Bai, Hui-Liang Shen
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 2D deformable attention to feature lifting (ablation) | SGCDet baseline · ScanNet | mAP@0.25: 56.0 -> 56.2 | No clear effect |
| add 3D deformable attention to feature lifting (ablation) | SGCDet baseline · ScanNet | mAP@0.25: 56.0 -> 59.5 | Improved |
| add multi-view attention on top of 3D deformable attention (ablation) | SGCDet · ScanNet | mAP@0.25: 59.5 -> 61.2 | Improved |
| add coarse-to-fine sparse refinement with 25% selection (ablation) | SGCDet · ScanNet | inference memory (GB): 22 -> 13 | Improved |
| add coarse-to-fine sparse refinement with 25% selection (ablation) | SGCDet · ScanNet | mAP@0.25: 61.0 -> 61.2 | No clear effect |
| reduce refinement selection ratio to 10% (ablation) | SGCDet · ScanNet | mAP@0.25: 61.2 -> 57.0 | Hurt |
| add occupancy loss (ablation) | SGCDet · ScanNet | mAP@0.25: 54.5 -> 61.2 | Improved |
| add DepthNet monocular branch (ablation) | SGCDet · ScanNet | mAP@0.25: 59.6 -> 61.2 | Improved |
| add DepthNet multi-view branch (ablation) | SGCDet · ScanNet | mAP@0.25: 57.7 -> 61.2 | Improved |
| add depth supervision (ablation) | SGCDet · ScanNet | mAP@0.25: 61.2 -> 62.2 | Improved |
| use ground-truth depth as input (ablation) | SGCDet · ScanNet | mAP@0.25: 61.2 -> 64.3 | Improved |
| train with noisy and incomplete box labels (ablation) | SGCDet vs ImGeoNet · ScanNet | mAP@0.50: 35.2 -> 33.6 | Mixed |
Closest papers by meaning
- MVSDet: Multi-View Indoor 3D Object Detection via Efficient Plane Sweeps (NeurIPS 2024 · similarity 0.87)
- SPGroup3D: Superpoint Grouping Network for Indoor 3D Object Detection (AAAI 2024 · similarity 0.86)
- Uncertainty Meets Diversity: A Comprehensive Active Learning Framework for Indoor 3D Object Detection (CVPR 2025 · similarity 0.86)
- Cubify Anything: Scaling Indoor 3D Object Detection (CVPR 2025 · similarity 0.86)
- GO-N3RDet: Geometry Optimized NeRF-enhanced 3D Object Detector (CVPR 2025 · similarity 0.85)
- CN-RMA: Combined Network with Ray Marching Aggregation for 3D Indoor Object Detection from Multi-view Images (CVPR 2024 · similarity 0.85)
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.
