Research map / 3D vision / Human pose estimation
TokenPose: Learning Keypoint Tokens for Human Pose Estimation
Yanjie Li, Shoukui Zhang, Zhicheng Wang, Sen Yang, Wankou Yang, Shu-Tao Xia, Erjin Zhou
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 2D sine position embedding (vs none) (ablation) | TokenPose-S-v1 · COCO val2017 | AP: 67.0 -> 72.5 | Improved |
| use 2D sine instead of learnable position embedding (ablation) | TokenPose-S-v1 · COCO val2017 | AP: 71.4 -> 72.5 | Improved |
| add learnable position embedding (vs none) (ablation) | TokenPose-S-v1 · COCO val2017 | AP: 67.0 -> 71.4 | Improved |
| add keypoint token fusion (ablation) | TokenPose-L+/D12 · COCO val2017 | AP: 75.3 -> 75.5 | Improved |
| add keypoint token fusion (ablation) | TokenPose-S · COCO val2017 | AP: 73.5 -> 72.6 | Hurt |
| increase depth from 12 to 24 layers (ablation) | TokenPose-L · COCO val2017 | AP: 75.3 -> 75.8 | Improved |
| increase embedding width from 192 to 384 (12 layers) (ablation) | TokenPose-L · COCO val2017 | AP: 75.3 -> 75.5 | Improved |
| replace SimpleBaseline-Res50 CNN with TokenPose-S-v1 | SimpleBaseline-Res50 · COCO val2017 | AP: 70.4 -> 72.5 | Improved |
| replace stem-net with HRNet-W32 backbone feeding tokens (TokenPose-B vs HRNet-W32) | HRNet-W32 · COCO val2017 | AP: 74.4 -> 74.7 | No clear effect |
| use pure Transformer without any convolution | TokenPose-T · COCO val2017 | AP: 65.6 vs 72.5 | Hurt |
| increase input resolution from 256x192 to 384x288 | TokenPose-L/D24 · COCO test-dev2017 | AP: 75.1 -> 75.9 | Improved |
| use patch size 2x2 instead of 4x3 (S-v2 vs S-v1) | TokenPose-S · COCO val2017 | AP: 72.5 -> 73.5 | Improved |
Closest papers by meaning
- TokenHMR: Advancing Human Mesh Recovery with a Tokenized Pose Representation (CVPR 2024 · similarity 0.88)
- ProbPose: A Probabilistic Approach to 2D Human Pose Estimation (CVPR 2025 · similarity 0.87)
- SDPose: Tokenized Pose Estimation via Circulation-Guide Self-Distillation (CVPR 2024 · similarity 0.86)
- FlexPose: Pose Distribution Adaptation with Limited Guidance (AAAI 2025 · similarity 0.86)
- Hourglass Tokenizer for Efficient Transformer-Based 3D Human Pose Estimation (CVPR 2024 · similarity 0.85)
- PoseBH: Prototypical Multi-Dataset Training Beyond Human Pose Estimation (CVPR 2025 · 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.
