Research map / Point clouds / Point cloud classification
HPLFlowNet: Hierarchical Permutohedral Lattice FlowNet for Scene Flow Estimation on Large-Scale 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 the HPLFlowNet architecture instead of Original BCL | Scene flow estimation · FlyingThings3D | EPE3D: 0.1111 -> 0.0804 | Helped |
| Use the HPLFlowNet architecture instead of Original BCL | Scene flow estimation · KITTI Scene Flow 2015 | EPE3D: 0.1729 -> 0.1169 | Helped |
| Use HPLFlowNet instead of FlowNet3D | Scene flow estimation · FlyingThings3D | EPE3D: 0.1136 -> 0.0804 | Helped |
| Use HPLFlowNet instead of FlowNet3D | Scene flow estimation · KITTI Scene Flow 2015 | EPE3D: 0.1767 -> 0.1169 | Helped |
| Use HPLFlowNet rather than FlowNet3D | Scene flow estimation · FlyingThings3D, 8,192 points | runtime (ms): 130.8 -> 98.4 | Helped |
| Add skip links between downsampling and upsampling stages (ablation) | HPLFlowNet · FlyingThings3D | EPE3D: 0.3149 -> 0.0804 | Helped |
| Use CorrBCLs at multiple scales instead of only the last CorrBCL (ablation) | HPLFlowNet · FlyingThings3D | EPE3D: 0.3698 -> 0.0804 | Helped |
| Use the proposed density normalization instead of original BCL normalization (ablation) | HPLFlowNet · FlyingThings3D | EPE3D: 0.6583 -> 0.0804 | Helped |
| Use concatenation rather than elementwise multiplication for patch correlation, retaining links between CorrBCLs (ablation) | HPLFlowNet · FlyingThings3D | EPE3D: 0.0948 -> 0.0804 | Helped |
| Add relative positions to input signals (ablation) | HPLFlowNet · FlyingThings3D | EPE3D: 0.0989 -> 0.0804 | Helped |
| Add density normalization (ablation) | HPLFlowNet · FlyingThings3D | EPE3D at 8,192 points: 0.0790 -> 0.0804 | Hurt |
| Add density normalization (ablation) | HPLFlowNet · FlyingThings3D | EPE3D at 65,536 points: 0.1267 -> 0.0772 | Helped |
Closest papers by meaning
- FlowNet3D: Learning Scene Flow in 3D Point Clouds (CVPR 2019 · similarity 0.89)
- PointFlowNet: Learning Representations for Rigid Motion Estimation From Point Clouds (CVPR 2019 · similarity 0.89)
- SCTN: Sparse Convolution-Transformer Network for Scene Flow Estimation (AAAI 2022 · similarity 0.88)
- HCRF-Flow: Scene Flow From Point Clouds With Continuous High-Order CRFs and Position-Aware Flow Embedding (CVPR 2021 · similarity 0.88)
- FlowMamba: Learning Point Cloud Scene Flow with Global Motion Propagation (AAAI 2025 · similarity 0.86)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · similarity 0.86)
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.