Research map / Point clouds / Segmentation and detection
SRGCD: Stability-Driven Region Growth Framework for 3D Change Detection
Yue Wu, Tao Peng, Yongzhe Yuan, Kaiyuan Feng, Hao Li, Maoguo Gong, Qiguang Miao, Wenping Ma
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 MGCP seeds and DynMask together to the region growth framework (ablation) | SRGCD · Urb3DCD-V1 | mIoU (%): 89.41 -> 94.11 | Improved |
| add MGCP seeds (without DynMask) (ablation) | SRGCD · Urb3DCD-V1 | mIoU (%): 89.41 -> 91.28 | Improved |
| add DynMask on top of MGCP (ablation) | SRGCD · Urb3DCD-V1 | mIoU (%): 91.28 -> 94.11 | Improved |
| add MGCP when only DynMask is used (ablation) | SRGCD · Urb3DCD-V1 | mIoU (%): 90.63 -> 94.11 | Improved |
| use SGA one-way attention instead of standard self-attention (ablation) | SRGCD · Urb3DCD-V1 | mIoU (%): 92.36 -> 94.11 | Improved |
| add Stability-Adaptive Propagation (SAP) (ablation) | SRGCD · Urb3DCD-V1 | mIoU (%): 93.17 -> 94.11 | Improved |
| add Stability-Proportional Temperature (SPT) (ablation) | SRGCD · Urb3DCD-V1 | mIoU (%): 93.25 -> 94.11 | Improved |
| extend region growth from Layer 0 only to all five levels (ablation) | SRGCD · Urb3DCD-V1 | mIoU (%): 90.92 -> 94.11 | Improved |
| extend region growth from Layers 0-3 to Layers 0-4 (ablation) | SRGCD · Urb3DCD-V1 | mIoU (%): 93.67 -> 94.11 | Improved |
| replace one-step segmentation decoder with stability-driven region growth | SiameseKPConv backbone · HKCD | mIoU (%): 73.73 -> 78.79 | Improved |
| replace one-step segmentation decoder with stability-driven region growth | SiameseKPConv backbone · Urb3DCD-V1 | mIoU (%): 89.89 -> 94.11 | Improved |
Closest papers by meaning
- MUCD: Unsupervised Point Cloud Change Detection via Masked Consistency (AAAI 2025 · similarity 0.87)
- CL3D: Unsupervised Domain Adaptation for Cross-LiDAR 3D Detection (AAAI 2023 · similarity 0.86)
- STGC-NeRF: Spatial-Temporal Geometric Consistency for LiDAR Neural Radiance Fields in Dynamic Scenes (AAAI 2025 · similarity 0.86)
- MOS: Model Synergy for Test-Time Adaptation on LiDAR-Based 3D Object Detection (ICLR 2025 · similarity 0.85)
- An Empirical Study of the Generalization Ability of Lidar 3D Object Detectors to Unseen Domains (CVPR 2024 · similarity 0.85)
- GBlobs: Explicit Local Structure via Gaussian Blobs for Improved Cross-Domain LiDAR-based 3D Object Detection (CVPR 2025 · 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.
