Research map / Point clouds / Point cloud classification
milliFlow: Scene Flow Estimation on mmWave Radar Point Cloud for Human Motion Sensing
Fangqiang Ding, Zhen Luo, Peijun Zhao, Chris Xiaoxuan Lu
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 temporal propagation (GRU on global vector) (ablation) | milliFlow · milliFlow in-set test | EPE3D All (m): 0.053 -> 0.046 | Improved |
| add global feature aggregation with attention (ablation) | milliFlow · milliFlow in-set test | EPE3D All (m): 0.061 -> 0.053 | Improved |
| add context features to flow embedding (ablation) | milliFlow · milliFlow in-set test | EPE3D All (m): 0.071 -> 0.061 | Improved |
| add constrained scene flow regression (clamp per-axis flow) (ablation) | milliFlow · milliFlow in-set test | EPE3D All (m): 0.083 -> 0.071 | Improved |
| use milliFlow instead of strongest baseline RaFlow | radar scene flow · milliFlow in-set test | EPE3D All (m): 0.107 -> 0.046 | Improved |
| feed estimated scene flow as per-point features (S1) to HAR network | HAR base network · Human activity recognition | overall accuracy (%): 47.32 -> 57.88 | Improved |
| use scene flow latent features (S2) for HAR | RadHAR · Human activity recognition | overall accuracy (%): 44.65 -> 50.53 | Improved |
| add scene flow as point features (S1) to human parsing | human parsing network · Human parsing | mIoU (%): 49.09 -> 52.72 | Improved |
| use latent scene flow features (S2) instead of explicit S1 for human parsing (ablation) | human parsing network · Human parsing | mIoU (%): 52.72 -> 51.04 | Hurt |
| track body parts over longer horizons with scene flow plus Kabsch | human body part tracking · HBPT | mJE (m) | Hurt |
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.
- PV-RAFT — PV-RAFT: Point-Voxel Correlation Fields for Scene Flow Estimation of Point Clouds (CVPR 2021) · better on 1 of 1 shared result
- FlowNet3D — FlowNet3D: Learning Scene Flow in 3D Point Clouds (CVPR 2019) · better on 1 of 1 shared result
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
- MiliPoint: A Point Cloud Dataset for mmWave Radar (NeurIPS 2023 · similarity 0.90)
- mmWaveFlow: Unified Enhancement and Generation of mmWave Human Point Clouds (CVPR 2026 · similarity 0.86)
- DAP: Doppler-aware Point Network for Heterogeneous mmWave Action Recognition (ECCV 2026 · similarity 0.86)
- MVDoppler-Pose: Multi-Modal Multi-View mmWave Sensing for Long-Distance Self-Occluded Human Walking Pose Estimation (CVPR 2025 · similarity 0.84)
- FlowMamba: Learning Point Cloud Scene Flow with Global Motion Propagation (AAAI 2025 · similarity 0.81)
- RadarMOSEVE: A Spatial-Temporal Transformer Network for Radar-Only Moving Object Segmentation and Ego-Velocity Estimation (AAAI 2024 · similarity 0.81)
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.
