Research map / 3D vision / Human pose estimation
AdaFuse: Adaptive Multiview Fusion for Accurate Human Pose Estimation in the Wild
Zhe Zhang, Chunyu Wang, Weichao Qiu, Wenhu Qin, Wenjun Zeng
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 multiview heatmap fusion (HeuristicFuse) | NoFuse · Human3.6M | 3D MPJPE (mm): 22.9 -> 21.0 | Improved |
| add adaptive fusion weights from appearance and geometry (AdaFuse) | HeuristicFuse · Human3.6M | 3D MPJPE (mm): 21.0 -> 19.5 | Improved |
| replace max-heatmap-score weights with learned weights | ScoreFuse · Human3.6M | 3D MPJPE (mm): 20.1 -> 19.5 | Improved |
| add geometry embedding to appearance embedding (ablation) | AdaFuse · Human3.6M | 3D MPJPE (mm): 20.3 -> 19.5 | Improved |
| add appearance embedding to geometry embedding (ablation) | AdaFuse · Human3.6M | 3D MPJPE (mm): 19.9 -> 19.5 | Improved |
| use multiview fusion instead of RANSAC outlier removal | RANSAC · Human3.6M | 3D MPJPE (mm): 21.8 -> 19.5 | Improved |
| add AdaFuse fusion | NoFuse · Occlusion-Person (occluded joints) | 2D PCKh@0.5 avg: 30.9 -> 95.5 | Improved |
| add AdaFuse fusion | NoFuse · Occlusion-Person | 3D MPJPE (mm): 48.1 -> 12.6 | Improved |
| add AdaFuse fusion | NoFuse · Occlusion-Person (joints occluded in 4 of 8 views) | 3D MPJPE (mm): 82.6 -> 21.7 | Improved |
| add AdaFuse fusion to joints visible in all views | NoFuse · Occlusion-Person (joints visible in all views) | 3D MPJPE (mm): 13.0 -> 10.8 | Improved |
| add AdaFuse fusion | NoFuse · Total Capture | 3D MPJPE (mm) mean: 29.4 -> 19.2 | Improved |
| add AdaFuse fusion to hip and head joints | NoFuse · Human3.6M | 3D MPJPE (mm): 17.8 -> 18.3 (Hip) | Hurt |
Later papers that compared with this method
2 later accepted papers report AdaFuse as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- PoseIRM: Enhance 3D Human Pose Estimation on Unseen Camera Settings via Invariant Risk Minimization (CVPR 2024 · better on 0 of 1 shared result)
- UPose3D: Uncertainty-Aware 3D Human Pose Estimation with Cross-View and Temporal Cues (ECCV 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
- MetaFuse: A Pre-trained Fusion Model for Human Pose Estimation (CVPR 2020 · similarity 0.92)
- Explicit Occlusion Reasoning for Multi-Person 3D Human Pose Estimation (ECCV 2022 · similarity 0.89)
- ODHSR: Online Dense 3D Reconstruction of Humans and Scenes from Monocular Videos (CVPR 2025 · similarity 0.88)
- FusionFormer: A Concise Unified Feature Fusion Transformer for 3D Pose Estimation (AAAI 2024 · similarity 0.87)
- $\text{Di}^2\text{Pose}$: Discrete Diffusion Model for Occluded 3D Human Pose Estimation (NeurIPS 2024 · similarity 0.86)
- PoseIRM: Enhance 3D Human Pose Estimation on Unseen Camera Settings via Invariant Risk Minimization (CVPR 2024 · similarity 0.86)
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.
