Research map / 3D vision / Human pose estimation
3D Human Pose Estimation Using Möbius Graph Convolutional Networks
Niloofar Azizi, Horst Possegger, Emanuele Rodolà, Horst Bischof
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 |
|---|---|---|---|
| use Mobius-transformed complex spectral filters instead of real Chebychev spectral GCN | MobiusGCN · Human3.6M Protocol #1 (2D GT input) | MPJPE (mm): 110.6 -> 40.0 | Improved |
| MobiusGCN 0.04M vs SemGCN 0.43M with 2D GT input | MobiusGCN · Human3.6M Protocol #1 (2D GT input) | MPJPE (mm): 43.8 -> 40.0 | Improved |
| MobiusGCN 0.04M vs SemGCN 0.43M with 2D HG input | MobiusGCN · Human3.6M Protocol #1 (2D HG input) | MPJPE (mm): 60.8 -> 54.2 | Improved |
| MobiusGCN 0.16M vs light GraphSH 0.44M | MobiusGCN · Human3.6M Protocol #1 (2D GT input) | MPJPE (mm): 39.2 -> 36.2 | Improved |
| increase channels from 64 to 128 (0.04M to 0.16M parameters), 2D GT input | MobiusGCN · Human3.6M Protocol #1 (2D GT input) | MPJPE (mm): 40.0 -> 36.2 | Improved |
| increase channels from 64 to 128 (0.04M to 0.16M parameters), 2D HG input | MobiusGCN · Human3.6M Protocol #1 (2D HG input) | MPJPE (mm): 54.2 -> 52.1 | Improved |
| MobiusGCN 0.16M vs GraphSH 3.7M, 2D HG input | MobiusGCN · Human3.6M Protocol #1 (2D HG input) | MPJPE (mm): 51.9 -> 52.1 | No clear effect |
| increase training subjects from S1 to S1+S5 (ablation) | MobiusGCN · Human3.6M training subjects (2D GT input) | MPJPE (mm): 62.3 -> 47.9 | Improved |
| increase training subjects from S1+S5 to S1+S5+S6 (ablation) | MobiusGCN · Human3.6M training subjects (2D GT input) | MPJPE (mm): 47.9 -> 43.1 | Improved |
| MobiusGCN vs GraphSH on MPI-INF-3DHP outdoor scene | MobiusGCN · MPI-INF-3DHP Outdoor | 3D PCK: 75.2 -> 83.1 | Improved |
| MobiusGCN vs GraphSH on MPI-INF-3DHP GS and noGS scenes | MobiusGCN · MPI-INF-3DHP GS / noGS | 3D PCK: 81.5 -> 79.2 (GS); 81.7 -> 77.3 (noGS) | Hurt |
| MobiusGCN trained on one subject vs semi-supervised methods using unlabeled data, 2D HG input | MobiusGCN · Human3.6M semi-supervised comparison (2D HG input) | MPJPE (mm): 88.8 -> 82.3 | Improved |
Closest papers by meaning
- Modulated Graph Convolutional Network for 3D Human Pose Estimation (ICCV 2021 · similarity 0.90)
- Locally Connected Network for Monocular 3D Human Pose Estimation (TPAMI 2020 · similarity 0.89)
- Semantic Graph Convolutional Networks for 3D Human Pose Regression (CVPR 2019 · similarity 0.89)
- Optimizing Network Structure for 3D Human Pose Estimation (ICCV 2019 · similarity 0.88)
- Pose2Mesh: Graph Convolutional Network for 3D Human Pose and Mesh Recovery from a 2D Human Pose (ECCV 2020 · similarity 0.88)
- Deep Semantic Graph Transformer for Multi-View 3D Human 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.
