Research map / Point clouds / Point cloud classification
What Matters for 3D Scene Flow Network
Guangming Wang, Yunzhe Hu, Zhe Liu, Yiyang Zhou, Masayoshi Tomizuka, Wei Zhan, Hesheng Wang
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 backward validation vector (ablation) | all-to-all flow embedding · FlyingThings3D (Gu et al., trained on quarter) | EPE3D: 0.0332 -> 0.0317 | Improved |
| add all-to-all mechanism (vs K nearest neighbours) (ablation) | flow embedding · FlyingThings3D (Gu et al., trained on quarter) | EPE3D: 0.0349 -> 0.0332 | Improved |
| use concatenated features instead of product similarity (ablation) | point similarity · FlyingThings3D (Gu et al., trained on quarter) | EPE3D: 0.0356 -> 0.0317 | Improved |
| use concatenated features instead of cosine similarity (ablation) | point similarity · FlyingThings3D (Gu et al., trained on quarter) | EPE3D: 0.0370 -> 0.0317 | Improved |
| use concatenated features instead of normalized product (ablation) | point similarity · FlyingThings3D (Gu et al., trained on quarter) | EPE3D: 0.0339 -> 0.0317 | Improved |
| use MLP predictor instead of GRU (ablation) | scene flow predictor · FlyingThings3D (Gu et al., trained on quarter) | EPE3D: 0.0350 -> 0.0317 | Improved |
| add up-sampled flow embedding to predictor input (ablation) | scene flow predictor · FlyingThings3D (Gu et al., trained on quarter) | EPE3D: 0.0380 -> 0.0317 | Improved |
| add PC1 point features to predictor input (ablation) | scene flow predictor · FlyingThings3D (Gu et al., trained on quarter) | EPE3D: 0.0333 -> 0.0317 | Improved |
| add coarse flow to predictor input (ablation) | scene flow predictor · FlyingThings3D (Gu et al., trained on quarter) | EPE3D: 0.0323 -> 0.0317 | Improved |
| add dense flow feature to predictor input (ablation) | scene flow predictor · FlyingThings3D (Gu et al., trained on quarter) | EPE3D: 0.0327 -> 0.0317 | Improved |
| add finest-level flow refinement layer instead of interpolation (2048 points) (ablation) | flow refinement · FlyingThings3D (Gu et al., trained on quarter) | EPE3D: 0.0359 -> 0.0317 | Improved |
| estimate 2048 points via refinement instead of 8192 points via interpolation (ablation) | flow refinement · FlyingThings3D (Gu et al., trained on quarter) | EPE3D: 0.0332 -> 0.0317 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- FESTA — FESTA: Flow Estimation via Spatial-Temporal Attention for Scene Point Clouds (CVPR 2021) · better on 2 of 2 shared results
- FlowNet3D — FlowNet3D: Learning Scene Flow in 3D Point Clouds (CVPR 2019) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Bi-PointFlowNet: Bidirectional Learning for Point Cloud Based Scene Flow Estimation (ECCV 2022 · similarity 0.89)
- PointFlowNet: Learning Representations for Rigid Motion Estimation From Point Clouds (CVPR 2019 · similarity 0.89)
- FlowNet3D: Learning Scene Flow in 3D Point Clouds (CVPR 2019 · similarity 0.89)
- DifFlow3D: Toward Robust Uncertainty-Aware Scene Flow Estimation with Iterative Diffusion-Based Refinement (CVPR 2024 · similarity 0.87)
- FlowMamba: Learning Point Cloud Scene Flow with Global Motion Propagation (AAAI 2025 · similarity 0.87)
- DeltaFlow: An Efficient Multi-frame Scene Flow Estimation Method (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.
