Research map / 3D vision / Human pose estimation
M4Human: A Large-Scale Multimodal mmWave Radar Benchmark for Human Mesh Reconstruction
Junqiao Fan, Yunjiao Zhou, Yizhuo Yang, Xinyuan Cui, Jiarui Zhang, Lihua Xie, Jianfei Yang, Chris Xiaoxuan Lu, Fangqiang Ding
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 |
|---|---|---|---|
| train on 100% instead of 25% of the data (ablation) | RT-Mesh · M4Human S2 cross-subject | MVE (mm): 161.0 -> 135.1 | Improved |
| train on 100% instead of 25% of the data (ablation) | RT-Mesh · M4Human S3 cross-action | MVE (mm): 174.9 -> 143.1 | Improved |
| use raw radar tensor instead of radar point cloud | P4Transformer vs RT-Mesh · M4Human ALL, S1 | MVE (mm): 90.4 -> 90.9 | No clear effect |
| use raw radar tensor instead of radar point cloud | RT-Mesh vs P4Transformer · M4Human ALL, S2 cross-subject | MVE (mm): 140.8 -> 135.1 | Improved |
| fuse RPC and RT | radar HMR · M4Human ALL, S1 | MVE (mm): 90.9 -> 84.3 | Improved |
| fuse RPC and RT | radar HMR · M4Human ALL, S2 | MVE (mm): 135.1 -> 135.2 | No clear effect |
| fuse RGB with RT | RT-Mesh · M4Human ALL, S2 | MVE (mm): 135.1 -> 112.5 | Improved |
| fuse RGB with RT | RT-Mesh · M4Human ALL, S1 | translation error (mm): 47.6 -> 51.6 | Hurt |
| fuse depth with RT | RT-Mesh · M4Human ALL, S1 | translation error (mm): 47.6 -> 36.0 | Improved |
| radar RT vs RGB for global translation | RT-Mesh vs RGB · M4Human ALL, S1 | translation error (mm): 54.4 -> 47.6 | Improved |
Closest papers by meaning
- mmPred: Radar-based Human Motion Prediction in the Dark (AAAI 2026 · similarity 0.87)
- SAT-HMR: Real-Time Multi-Person 3D Mesh Estimation via Scale-Adaptive Tokens (CVPR 2025 · similarity 0.86)
- MVDoppler-Pose: Multi-Modal Multi-View mmWave Sensing for Long-Distance Self-Occluded Human Walking Pose Estimation (CVPR 2025 · similarity 0.86)
- MMVR: Millimeter-wave Multi-View Radar Dataset and Benchmark for Indoor Perception (ECCV 2024 · similarity 0.86)
- MeshPose: Unifying DensePose and 3D Body Mesh Reconstruction (CVPR 2024 · similarity 0.85)
- Reconstructing People, Places, and Cameras (CVPR 2025 · similarity 0.85)
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.
