Research map / 3D vision / Human pose estimation
H 2 OT: Hierarchical Hourglass Tokenizer for Efficient Video Pose Transformers
Wenhao Li, Mengyuan Liu, Hong Liu, Pichao Wang, Shijian Lu, Nicu Sebe
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 H2OT hierarchical pruning and recovering | MixSTE (seq2seq) · Human3.6M | MPJPE (mm): 40.9 -> 40.5 | Improved |
| add H2OT token pruning | MixSTE (seq2seq) · Human3.6M | FLOPs (G): 277.25 -> 118.23 | Improved |
| add H2OT token pruning | MotionBERT (seq2seq) · Human3.6M | MPJPE (mm): 39.8 -> 39.9 | No clear effect |
| replace HoT with H2OT (hierarchical, TPS and TRI) | MixSTE · Human3.6M | MPJPE (mm): 41.0 -> 40.5 | Improved |
| use sequence-sampler pruning with interpolation recovering instead of clustering with attention recovering (ablation) | MixSTE · Human3.6M | FPS per frame: 56 -> 66 | Improved |
| use motion-based pruning (TPMo) instead of clustering (TPC) (ablation) | MixSTE · Human3.6M | MPJPE (mm): 41.0 -> 42.8 | Hurt |
| use attention-based pruning (TPA) instead of clustering (TPC) (ablation) | MixSTE · Human3.6M | MPJPE (mm): 41.0 -> 42.1 | Hurt |
| add hierarchical second pruning stage r=[121,81] over single stage r=81 (ablation) | MixSTE · Human3.6M | MPJPE (mm): 41.0 -> 40.5 | Improved |
| prune more tokens (r=[121,27] instead of [121,81]) (ablation) | MixSTE · Human3.6M | MPJPE (mm): 40.5 -> 41.3 | Hurt |
| increase number of recovered tokens f from 9 to 243 (ablation) | MixSTE (seq2frame) · Human3.6M | MPJPE (mm): 41.2 -> 40.4 | Improved |
| add H2OT under 12.5 FPS input (ablation) | MixSTE · Human3.6M | MPJPE (mm): 39.2 -> 40.8 | Hurt |
| add H2OT under 50 FPS input (ablation) | MixSTE · Human3.6M | MPJPE (mm): 40.9 -> 40.5 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- MixSTE — MixSTE: Seq2seq Mixed Spatio-Temporal Encoder for 3D Human Pose Estimation in Video (CVPR 2022) · better on 4 of 4 shared results
- DiffPose — DiffPose: Toward More Reliable 3D Pose Estimation (CVPR 2023) · better on 1 of 1 shared result
- MHFormer — MHFormer: Multi-Hypothesis Transformer for 3D Human Pose Estimation (CVPR 2022) · better on 1 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
- Hourglass Tokenizer for Efficient Transformer-Based 3D Human Pose Estimation (CVPR 2024 · similarity 0.96)
- TokenHMR: Advancing Human Mesh Recovery with a Tokenized Pose Representation (CVPR 2024 · similarity 0.88)
- HiPART: Hierarchical Pose AutoRegressive Transformer for Occluded 3D Human Pose Estimation (CVPR 2025 · similarity 0.87)
- SDPose: Tokenized Pose Estimation via Circulation-Guide Self-Distillation (CVPR 2024 · similarity 0.87)
- Efficient Long Video Tokenization via Coordinate-based Patch Reconstruction (CVPR 2025 · similarity 0.86)
- ExtPose: Robust and Coherent Pose Estimation by Extending ViTs (ICML 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.
