Research map / 3D vision / Human pose estimation
Geometry-Contrastive Transformer for Generalized 3D Pose Transfer
Haoyu Chen, Hao Tang, Zitong Yu, Nicu Sebe, Guoying Zhao
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 GC-Transformer instead of NPT | 3D pose transfer · SMG-3D seen | PMD (× 10^-4): 62.1 -> 30.7 | Improved |
| Use GC-Transformer instead of NPT | 3D pose transfer · SMG-3D unseen | PMD (× 10^-4): 94.6 -> 52.8 | Improved |
| Use GC-Transformer instead of NPT | 3D pose transfer · SMPL-NPT seen | PMD (× 10^-4): 1.1 -> 0.6 | Improved |
| Use GC-Transformer instead of NPT | 3D pose transfer · SMPL-NPT unseen | PMD (× 10^-4): 9.3 -> 4.0 | Improved |
| Use GC-Transformer rather than geodesic-preservation LIMP-Geo | 3D pose transfer · FAUST | Disentanglement Error: 3.48 -> 0.11 | Improved |
| Apply LIR to the model | GC-Transformer · SMG-3D cross-dataset from SMPL-NPT | PMD (× 10^-4): 178.7 -> 79.2 | Improved |
| Apply LIR to the model | NPT · SMG-3D cross-dataset from SMPL-NPT | PMD (× 10^-4): 240.1 -> 121.4 | Improved |
| Apply LIR to the model | NPT-MP · SMG-3D cross-dataset from SMPL-NPT | PMD (× 10^-4): 321.4 -> 132.3 | Improved |
| Train on semi-synthesized SMG-3D rather than synthesized SMPL-NPT | NPT · SMG-3D | PMD (× 10^-4): 321.4 -> 120.3 | Improved |
| Train on semi-synthesized SMG-3D rather than synthesized SMPL-NPT | GC-Transformer · SMG-3D | PMD (× 10^-4): 178.7 -> 52.8 | Improved |
| Increase attention depth from 1 to 4 blocks (ablation) | GC-Transformer · SMPL-NPT seen-pose | PMD (× 10^-4): 1.4 -> 0.8 | Improved |
| Increase attention depth from 1 to 4 blocks (ablation) | GC-Transformer · SMPL-NPT unseen-pose | PMD (× 10^-4): 7.3 -> 4.2 | Improved |
Closest papers by meaning
- MAPConNet: Self-supervised 3D Pose Transfer with Mesh and Point Contrastive Learning (ICCV 2023 · similarity 0.90)
- Unsupervised 3D Pose Transfer With Cross Consistency and Dual Reconstruction (TPAMI 2023 · similarity 0.88)
- Towards Robust 3D Pose Transfer with Adversarial Learning (CVPR 2024 · similarity 0.86)
- Multiple View Geometry Transformers for 3D Human Pose Estimation (CVPR 2024 · similarity 0.86)
- Neural Pose Representation Learning for Generating and Transferring Non-Rigid Object Poses (NeurIPS 2024 · similarity 0.86)
- Not All Tasks Quantize Equally: Fisher-Guided Quantization for Visual Geometry Transformer (NeurIPS 2026 · 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.
