Research map / Point clouds / Point cloud classification
Open-Vocabulary Octree-Graph for 3D Scene Understanding
Zhigang Wang, Yifei Su, Chenhui Li, Dong Wang, Yan Huang, Xuelong Li, Bin Zhao
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 |
|---|---|---|---|
| use CGSM instead of frame-wise merging (ablation) | Octree-Graph · ScanNet semantic segmentation | mIoU: 0.323 -> 0.356 | Improved |
| use CGSM instead of global-wise merging (ablation) | Octree-Graph · ScanNet semantic segmentation | mIoU: 0.286 -> 0.356 | Improved |
| vary group interval I from 100 to 400 (ablation) | CGSM · ScanNet semantic segmentation | mIoU: 0.344 / 0.356 / 0.338 for I=100/200/400 | No clear effect |
| add semantic-guided under-segment filtering (ablation) | group-wise merging · ScanNet semantic segmentation | mIoU: 0.337 -> 0.346 | Improved |
| add threshold decay (ablation) | group-wise merging with filtering · ScanNet semantic segmentation | mIoU: 0.346 -> 0.356 | Improved |
| add semantic-guided under-segment filtering (ablation) | group-wise merging · ScanNet semantic segmentation | mAcc: 0.547 -> 0.557 | Improved |
| use IFA instead of plain average feature aggregation (ablation) | Octree-Graph · ScanNet semantic segmentation | mIoU: 0.338 -> 0.356 | Improved |
| use Top-5 instead of plain averaging for feature aggregation (ablation) | Octree-Graph · ScanNet semantic segmentation | mIoU: 0.338 -> 0.342 | Improved |
| use DBSCAN instead of plain averaging for feature aggregation (ablation) | Octree-Graph · ScanNet semantic segmentation | mIoU: 0.338 -> 0.345 | Improved |
| use adaptive-octree instead of classical octree at same depth (ablation) | Octree-Graph · Replica occupancy accuracy | mEOR: 0.0057 -> 0.0108 | Improved |
| use adaptive-octree instead of classical octree (ablation) | Octree-Graph · Replica storage | storage: 17.6KB -> 29.8KB | Hurt |
| use octree instead of point cloud (ablation) | scene representation · Replica storage | storage: 18.5MB -> 29.8KB | Improved |
Closest papers by meaning
- OpenLex3D: A Tiered Benchmark for Open-Vocabulary 3D Scene Representations (NeurIPS 2025 · similarity 0.90)
- Open3DSG: Open-Vocabulary 3D Scene Graphs from Point Clouds with Queryable Objects and Open-Set Relationships (CVPR 2024 · similarity 0.90)
- Cross-Modal and Uncertainty-Aware Agglomeration for Open-Vocabulary 3D Scene Understanding (CVPR 2025 · similarity 0.90)
- Open3DIS: Open-Vocabulary 3D Instance Segmentation with 2D Mask Guidance (CVPR 2024 · similarity 0.89)
- A Unified Framework for 3D Scene Understanding (NeurIPS 2024 · similarity 0.89)
- Masked Point-Entity Contrast for Open-Vocabulary 3D Scene Understanding (CVPR 2025 · 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.
