Research map / Point clouds / Registration and matching
mmWaveFlow: Unified Enhancement and Generation of mmWave Human Point Clouds
Chang Su, Beihong Jin, Qiwen Shi, Zhi 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 cross-modal latent alignment with shared queries (ablation) | mmWaveFlow · MM-Fi D2M | EMD (cm): 10.07 -> 8.49 | Improved |
| add origin-aware conditioning (OA-Flow) (ablation) | mmWaveFlow · mRI D2M | EMD (cm): 9.79 -> 9.22 | Improved |
| add OA-Flow (ablation) | mmWaveFlow · mmBody M2D | EMD (cm): 5.18 -> 4.76 | Improved |
| add latent alignment and OA-Flow together versus neither (ablation) | mmWaveFlow · MM-Fi D2M | EMD (cm): 13.48 -> 8.49 | Improved |
| add latent alignment on mRI D2M (ablation) | mmWaveFlow · mRI D2M | EMD (cm): 9.24 -> 9.22 | No clear effect |
| enhance mmWave inputs with mmWaveFlow before training mmDiff | mmDiff · mmBody human mesh recovery | MPJPE (cm): 7.97 -> 7.25 | Improved |
| enhance mmWave inputs with mmWaveFlow | P4Transformer · mmBody human mesh recovery | MPJPE (cm): 13.65 -> 12.59 | Improved |
| augment half real data with equal synthetic mmWave generated by mmWaveFlow | 3DInAction · MM-Fi action recognition | accuracy: 0.661 -> 0.790 | Improved |
| train 3DInAction on dense point clouds ground truth only (no mmWave) | 3DInAction · MM-Fi action recognition | accuracy: 0.156 | Hurt |
| use origin-aware flow with single-step versus multi-step integration | condition-free flow matching · latent transport | transport quality | Mixed |
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.
- Tiger — TIGER: Time-Varying Denoising Model for 3D Point Cloud Generation with Diffusion Process (CVPR 2024) · better on 6 of 6 shared results
- mmPoint — MM-Point: Multi-View Information-Enhanced Multi-Modal Self-Supervised 3D Point Cloud Understanding (AAAI 2024) · 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
- milliFlow: Scene Flow Estimation on mmWave Radar Point Cloud for Human Motion Sensing (ECCV 2024 · similarity 0.86)
- DAP: Doppler-aware Point Network for Heterogeneous mmWave Action Recognition (ECCV 2026 · similarity 0.86)
- Expanding mmWave Datasets for Human Pose Estimation with Unlabeled Data and LiDAR Datasets (CVPR 2026 · similarity 0.85)
- MITO: A Millimeter-Wave Dataset and Simulator for Non-Line-of-Sight Perception (NeurIPS 2026 · similarity 0.83)
- MVDoppler-Pose: Multi-Modal Multi-View mmWave Sensing for Long-Distance Self-Occluded Human Walking Pose Estimation (CVPR 2025 · similarity 0.83)
- Metric Flow Matching for Smooth Interpolations on the Data Manifold (NeurIPS 2024 · similarity 0.83)
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.
