Research map / Point clouds / Point cloud classification
HCRF-Flow: Scene Flow From Point Clouds With Continuous High-Order CRFs and Position-Aware Flow Embedding
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 |
|---|---|---|---|
| Add Con-HCRFs refinement to PAFE (ablation) | PAFE · FlyingThings3D test | Acc3DS: 78.90 -> 83.37 | Helped |
| Add Con-HCRFs refinement to PAFE-S (ablation) | PAFE-S · FlyingThings3D test | Acc3DS: 50.08 -> 56.29 | Helped |
| Add pairwise spatial-smoothness term (ablation) | PAFE-S · FlyingThings3D test | Acc3DS: 50.08 -> 50.24 | Helped |
| Replace rigid-motion high-order term with naive region-level average-flow smoothing (ablation) | PAFE-S with Con-HCRFs · FlyingThings3D test | Acc3DS: 50.08 -> 48.10 | Hurt |
| Add rigid-motion high-order term instead of naive regional smoothing (ablation) | PAFE-S with Unary+Pair · FlyingThings3D test | Acc3DS: 50.08 -> 54.51 | Helped |
| Add pseudo-pairing and position encoding to aggregation (ablation) | Single-scale baseline · FlyingThings3D test | Acc3DS: 41.63 -> 50.08 | Helped |
| Use pseudo-pairing rather than naive dynamic aggregation (ablation) | Single-scale flow embedding · FlyingThings3D test | Acc3DS: + 2.62 -> + 6.79 | Helped |
| Add pseudo-pairing and position encoding to aggregation (ablation) | Pyramid baseline · FlyingThings3D test | Acc3DS: 69.94 -> 78.90 | Helped |
| Jointly optimize Con-HCRFs with the PAFE module (ablation) | PAFE with Con-HCRFs · FlyingThings3D test | Acc3DS: 81.39 -> 83.37 | Helped |
| Add Con-HCRFs post-processing | FlowNet3D · KITTI Scene Flow 2015 | Acc3DS: 42.77 -> 46.90 | Helped |
| Add Con-HCRFs post-processing | FLOT · FlyingThings3D | Acc3DS: 73.20 -> 78.63 | Helped |
| Increase desired supervoxel size from 80 to 140 points (ablation) | HCRF-Flow · FlyingThings3D test | EPE3D: 0.0804 -> 0.0782 | Helped |
Closest papers by meaning
- FlowMamba: Learning Point Cloud Scene Flow with Global Motion Propagation (AAAI 2025 · similarity 0.88)
- Flow-NeRF: Joint Learning of Geometry, Poses, and Dense Flow within Unified Neural Representations (CVPR 2025 · similarity 0.88)
- ZeroFlow: Scalable Scene Flow via Distillation (ICLR 2024 · similarity 0.87)
- VoteFlow: Enforcing Local Rigidity in Self-Supervised Scene Flow (CVPR 2025 · similarity 0.87)
- ICP-Flow: LiDAR Scene Flow Estimation with ICP (CVPR 2024 · similarity 0.87)
- Rectified Point Flow: Generic Point Cloud Pose Estimation (NeurIPS 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.