Research map / Point clouds / Point cloud classification
Deformation and Correspondence Aware Unsupervised Synthetic-to-Real Scene Flow Estimation for Point Clouds
Check 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 |
|---|---|---|---|
| Use GTA-SF instead of FT3D as the synthetic training dataset | HPLFlowNet transferred to Waymo · Waymo | EPE3D: 0.3546 -> 0.1589; 0.2505 -> 0.1146; 0.3299 -> 0.1081; 0.2621 -> 0.0585 | Helped |
| Add deformation regularization (DR) to EPC (ablation) | Ours (E), HPLFlowNet · GTA-SF→Waymo | EPE: 0.0894 -> 0.0887 | Helped |
| Add correspondence refinement (CR) after EPC and DR (ablation) | Ours (E+D), HPLFlowNet · GTA-SF→Waymo, Lyft, and KITTI | EPE: 0.0887 -> 0.0683; 0.1454 -> 0.1277; 0.0748 -> 0.0464 | Helped |
| Add DR to EPC (ablation) | Ours (E), HPLFlowNet · FT3D→Waymo, Lyft, and KITTI | EPE: 0.2339 -> 0.2091; 0.7330 -> 0.5092; 0.1193 -> 0.0992 | Helped |
| Add CR after EPC and DR (ablation) | Ours (E+D), HPLFlowNet · FT3D→Waymo, Lyft, and KITTI | EPE: 0.2091 -> 0.1251; 0.5092 -> 0.4442; 0.0992 -> 0.0516 | Helped |
| Use an EMA mean-teacher rather than a student-identical teacher (ablation) | Ours, HPLFlowNet · GTA-SF→Waymo | EPE: 0.0768 -> 0.0683 | Helped |
| Use asymmetric rather than symmetric input transformations (ablation) | Ours, HPLFlowNet · GTA-SF→Waymo | EPE: 0.0988 -> 0.0683 | Helped |
| Use rotation-only asymmetric transformation instead of translation-only (ablation) | Ours, HPLFlowNet · GTA-SF→Waymo | EPE: 0.0719 -> 0.0683 | Helped |
| Use rotation-only rather than translation-plus-rotation augmentation (ablation) | Ours, HPLFlowNet · GTA-SF→Waymo | EPE: 0.0702 -> 0.0683 | Helped |
| Apply the proposed UDA framework to PV-RAFT | PV-RAFT · GTA-SF→Waymo | EPE: 0.0585 -> 0.0474 | Helped |
| Apply the proposed UDA framework to FLOT | FLOT · GTA-SF→Waymo | EPE: 0.1081 -> 0.0888 | Helped |
Closest papers by meaning
- 3DSFLabelling: Boosting 3D Scene Flow Estimation by Pseudo Auto-labelling (CVPR 2024 · similarity 0.89)
- Rectified Point Flow: Generic Point Cloud Pose Estimation (NeurIPS 2025 · similarity 0.88)
- ZeroFlow: Scalable Scene Flow via Distillation (ICLR 2024 · similarity 0.88)
- ICP-Flow: LiDAR Scene Flow Estimation with ICP (CVPR 2024 · similarity 0.88)
- Living Scenes: Multi-object Relocalization and Reconstruction in Changing 3D Environments (CVPR 2024 · similarity 0.87)
- Zero-Shot Monocular Scene Flow Estimation in the Wild (CVPR 2025 · similarity 0.87)
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.