Research map / 3D vision / Human pose estimation
High-Resolution Spatiotemporal Modeling with Global-Local State Space Models for Video-Based Human Pose Estimation
Runyang Feng, Hyung Jin Chang, Tze Ho Elden Tse, Boeun Kim, Yi Chang, Yixing Gao
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 Global Spatiotemporal Mamba to backbone (ablation) | GLSMamba-B · PoseTrack2017 val (ablation) | mAP: 74.2 -> 86.0 | Improved |
| add Local Refinement Mamba (ablation) | GLSMamba-B · PoseTrack2017 val (ablation) | mAP: 86.0 -> 86.9 | Improved |
| add space-wise scan to unified scan (ablation) | STS6D · PoseTrack2017 val (scan ablation) | mAP: 85.8 -> 86.5 | Improved |
| add time-wise scan to unified + space-wise (ablation) | STS6D · PoseTrack2017 val (scan ablation) | mAP: 86.5 -> 86.9 | Improved |
| add STMM instead of simple addition merging (ablation) | GSM · PoseTrack2017 val (scan ablation) | mAP: 86.2 -> 86.9 | Improved |
| use 1/4 resolution instead of 1/16 (ablation) | GLSMamba-B · PoseTrack2017 val (resolution study) | mAP: 85.7 -> 86.9 | Improved |
| use 1/8 resolution instead of 1/16 (ablation) | Transformer · PoseTrack2017 val (resolution study) | mAP: 84.2 -> 84.8 | Improved |
| use 1/4 resolution with Transformer (ablation) | Transformer · PoseTrack2017 val (resolution study) | mAP: OOM vs 138.9 GFLOPs | Hurt |
| GLSMamba-B instead of TDMI | TDMI · PoseTrack2017 val | mAP: 85.7 -> 86.9 | Improved |
| GLSMamba-B instead of DiffPose | DiffPose · PoseTrack2017 val | mAP: 86.4 -> 86.9 | Improved |
| GLSMamba-B instead of FAMI-Pose | FAMI-Pose · Sub-JHMDB | mAP: 96.0 -> 97.9 | Improved |
| GLSMamba-B instead of DeciWatch | DeciWatch · Sub-JHMDB | mAP: 98.8 -> 97.9 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DiffPose — DiffPose: Toward More Reliable 3D Pose Estimation (CVPR 2023) · better on 3 of 3 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
- PoseMamba: Monocular 3D Human Pose Estimation with Bidirectional Global-Local Spatio-Temporal State Space Model (AAAI 2025 · similarity 0.90)
- Video-Based Human Pose Regression via Decoupled Space-Time Aggregation (CVPR 2024 · similarity 0.88)
- SpatioTemporal Learning for Human Pose Estimation in Sparsely-Labeled Videos (AAAI 2025 · similarity 0.88)
- Spatial-Mamba: Effective Visual State Space Models via Structure-Aware State Fusion (ICLR 2025 · similarity 0.87)
- MV-SSM: Multi-View State Space Modeling for 3D Human Pose Estimation (CVPR 2025 · similarity 0.87)
- Pose Magic: Efficient and Temporally Consistent Human Pose Estimation with a Hybrid Mamba-GCN Network (AAAI 2025 · similarity 0.87)
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.
