Research map / 3D vision / Human pose estimation
GTPT: Group-based Token Pruning Transformer for Efficient Human Pose Estimation
Haonan Wang, Jie Liu, Jie Tang, Gangshan Wu, Bo Xu, Yanbing Chou, Yong Wang
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 |
|---|---|---|---|
| group keypoint tokens (head, upper body, lower body) (ablation) | GTPT-S · COCO-WholeBody val | Whole-body AP: 58.2 -> 59.1 | Improved |
| add per-group masking of visual tokens (ablation) | GTPT-S · COCO-WholeBody val | Whole-body AP: 59.1 -> 59.6 | Improved |
| use MHGA with shared sparse keypoint tokens instead of MHSA within groups (ablation) | GTPT-S · COCO-WholeBody val | Whole-body AP: 58.8 -> 59.6 | Improved |
| introduce keypoints coarse-to-fine (human-sparse-dense) instead of all dense at once (ablation) | GTPT-S · COCO-WholeBody val | GFLOPs / AP: 2.26 -> 1.99 GFLOPs; 59.8 -> 59.6 AP | Mixed |
| add global perceived loss to pruning (ablation) | GTPT-S with grouping and masking · COCO-WholeBody val | Whole-body AP: 59.2 -> 59.6 | Improved |
| prune visual tokens (ablation) | GTPT-S · COCO-WholeBody val | GFLOPs: 3.53 -> 1.99 (-44%) | Improved |
| add grouping and masking when pruning directly (ablation) | GTPT-S · COCO-WholeBody val | Whole-body AP: 56.6 -> 59.2 | Improved |
| add curriculum learning (body then whole-body) (ablation) | GTPT-S · COCO-WholeBody val | Whole-body AP: 56.5 -> 59.6 | Improved |
| token pruning on single-instance latency (ablation) | GTPT-S · COCO-WholeBody val | FPS: 93 -> 91 FPS | No clear effect |
| token pruning on batch throughput and memory (ablation) | GTPT-S · COCO-WholeBody val | throughput: 701 -> 1231 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- ZoomNAS — ZoomNAS: Searching for Whole-body Human Pose Estimation in the Wild (TPAMI 2022) · better on 0 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
- Transformers With Joint Tokens and Local-Global Attention for Efficient Human Pose Estimation (TPAMI 2026 · similarity 0.88)
- PPT: Token-Pruned Pose Transformer for Monocular and Multi-View Human Pose Estimation (ECCV 2022 · similarity 0.88)
- Hourglass Tokenizer for Efficient Transformer-Based 3D Human Pose Estimation (CVPR 2024 · similarity 0.88)
- SDPose: Tokenized Pose Estimation via Circulation-Guide Self-Distillation (CVPR 2024 · similarity 0.87)
- Optimizing Human Pose Estimation Through Focused Human and Joint Regions (AAAI 2025 · similarity 0.85)
- HiPART: Hierarchical Pose AutoRegressive Transformer for Occluded 3D 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.
