Research map / 3D vision / Human pose estimation
SMPLer: Taming Transformers for Monocular 3D Human Shape and Pose Estimation
Xiangyu Xu, Lijuan Liu, Shuicheng Yan
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 multi-scale features (ablation) | SMPLer · Human3.6M | MPJPE: 48.3 -> 47.0 | Improved |
| use SMPL-based target with decoupled attention instead of full attention and vertex-based (ablation) | Transformer · Human3.6M | MPJPE: 51.9 -> 47.0 | Improved |
| use SMPL-based target instead of vertex-based (ablation) | decoupled attention multi-scale · GPU memory | memory (G): 1.15 -> 0.44 | Improved |
| use decoupled instead of full attention (ablation) | vertex-based multi-scale · GPU memory | memory (G): 7.3 -> 1.15 | Improved |
| use separate projections per scale instead of concatenation (ablation) | SMPLer · Human3.6M | MPJPE: 48.5 -> 47.0 | Improved |
| add feature positional encoding (ablation) | SMPLer · Human3.6M | MPJPE: 50.4 -> 47.0 | Improved |
| use pooling-based FPE instead of all-scale FPE (ablation) | SMPLer · Human3.6M | MPJPE: 49.0 -> 47.0 | Improved |
| add joint-aware attention (ablation) | SMPLer · Human3.6M | MPJPE: 51.4 -> 47.0 | Improved |
| add relative positional encoding eta (ablation) | SMPLer · Human3.6M | MPJPE: 48.7 -> 47.0 | Improved |
| increase Transformer blocks from 1 to 3 (ablation) | SMPLer · Human3.6M | MPJPE: 48.6 -> 47.0 | Improved |
| increase units per block from 1 to 2 (ablation) | SMPLer · Human3.6M | MPJPE: 48.5 -> 47.0 | Improved |
| increase units per block from 2 to 3 (ablation) | SMPLer · Human3.6M | MPJPE: 47.0 -> 47.1 | No clear effect |
Later papers that compared with this method
1 later accepted paper reports SMPLer-X-L as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- MeshMamba: State Space Models for Articulated 3D Mesh Generation and Reconstruction (ICCV 2025 · better on 2 of 2 shared results)
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
- SMPLer-X: Scaling Up Expressive Human Pose and Shape Estimation (NeurIPS 2023 · similarity 0.89)
- Multiple View Geometry Transformers for 3D Human Pose Estimation (CVPR 2024 · similarity 0.87)
- Keep It SMPL: Automatic Estimation of 3D Human Pose and Shape from a Single Image (ECCV 2016 · similarity 0.87)
- Cross-Attention of Disentangled Modalities for 3D Human Mesh Recovery with Transformers (ECCV 2022 · similarity 0.87)
- 3D Human Pose Estimation With Spatial and Temporal Transformers (ICCV 2021 · similarity 0.86)
- HeatFormer: A Neural Optimizer for Multiview Human Mesh Recovery (CVPR 2025 · 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.
