Research map / 3D vision / Human pose estimation
From Sparse to Dense: Spatio-Temporal Fusion for Multi-View 3D Human Pose Estimation with DenseWarper
Ling Li, Changjie Chen, Yuyan Wang, Jiaqing Lyu, Kenglun Chang, Yiyun Chen, Zhidong Deng
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 spatial epipolar heatmap fusion (ablation) | DenseWarper baseline · Human3.6M (SimpleBaseline) | MPJPE (mm): 36.06 -> 31.54 | Improved |
| add Warper temporal fusion on top of spatial fusion (ablation) | DenseWarper · Human3.6M (SimpleBaseline) | MPJPE (mm): 31.54 -> 22.28 | Improved |
| add spatial epipolar heatmap fusion (ablation) | DenseWarper baseline · MPI-INF-3DHP | MPJPE (mm): 94.46 -> 88.63 | Improved |
| add Warper on top of spatial fusion (ablation) | DenseWarper · MPI-INF-3DHP | MPJPE (mm): 88.63 -> 65.89 | Improved |
| use sparse interleaved DenseWarper instead of Adafuse sparse | Adafuse · Human3.6M (GT 2D) | MPJPE (mm): 25.4 -> 21.3 | Improved |
| use DenseWarper sparse vs Adafuse full-frame | Adafuse · Human3.6M (GT 2D) | MPJPE (mm): 23.7 -> 21.3 | Improved |
| use DenseWarper vs interpolation (SLERP) | Adafuse · MPI-INF-3DHP | MPJPE (mm): 83.37 -> 65.89 | Improved |
| use sparse input instead of full-frame input | Adafuse · Human3.6M (GT 2D) | MPJPE (mm): 23.7 -> 25.4 | Hurt |
| use DenseWarper vs Volumetric full-frame | Volumetric · Human3.6M (SimpleBaseline) | MPJPE (mm): 26.7 -> 22.3 | Improved |
| use DenseWarper vs Sgraformer | Sgraformer · Human3.6M P-MPJPE | P-MPJPE (mm): 19.9 -> 19.4 | Improved |
| use DenseWarper vs Adafuse | Adafuse · Human3.6M (SimpleBaseline) latency | latency (ms): 96.028 -> 44.51 | Improved |
| single-view KTP-Former vs DenseWarper | KTP-Former · MPI-INF-3DHP | MPJPE (mm): 67.59 -> 65.89 | Improved |
Closest papers by meaning
- SVTformer: Spatial-View-Temporal Transformer for Multi-View 3D Human Pose Estimation (AAAI 2025 · similarity 0.89)
- TEMPO: Efficient Multi-View Pose Estimation, Tracking, and Forecasting (ICCV 2023 · similarity 0.89)
- MV-SSM: Multi-View State Space Modeling for 3D Human Pose Estimation (CVPR 2025 · similarity 0.88)
- Sparse-view Pose Estimation and Reconstruction via Analysis by Generative Synthesis (NeurIPS 2024 · similarity 0.87)
- ODHSR: Online Dense 3D Reconstruction of Humans and Scenes from Monocular Videos (CVPR 2025 · similarity 0.87)
- FusionFormer: A Concise Unified Feature Fusion Transformer for 3D Pose Estimation (AAAI 2024 · 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.
