Research map / 3D vision / Human pose estimation
Beyond Static Frames: Temporal Aggregate-and-Restore Vision Transformer for Human Pose Estimation
Hongwei Fang, Jiahang Cai, Xun Wang, Wenwu Yang
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 temporal aggregate-and-restore to ViTPose | ViTPose ViT-B · PoseTrack2017 val | mAP: 81.7 -> 84.0 | Improved |
| add temporal aggregate-and-restore to ViTPose | ViTPose ViT-H · PoseTrack2017 val | mAP: 84.7 -> 86.8 | Improved |
| joint-specific temporal modeling (JTA) instead of cross-attention over aux frames (ablation) | TAR-ViTPose ViT-B · PoseTrack2017 val | mAP: 82.8 -> 84.0 | Improved |
| generic self/cross-attention variants (ablation) | TAR-ViTPose ViT-B · PoseTrack2017 val | mAP: 82.2 / 82.6 / 82.8 vs baseline 81.7 | No clear effect |
| add Global Restoring Attention (GRA) (ablation) | TAR-ViTPose ViT-B · PoseTrack2017 val | mAP: 70.3 -> 84.0 | Improved |
| add mask-aware attention to JTA (ablation) | TAR-ViTPose ViT-B · PoseTrack2017 val | mAP: 82.6 -> 84.0 | Improved |
| use ground-truth boxes instead of detector boxes | TAR-ViTPose ViT-H · PoseTrack2017 val | mAP: 86.8 -> 90.3 | Improved |
| use ViT-S TAR-ViTPose over PoseWarper for speed | TAR-ViTPose · PoseTrack2017 runtime | FPS: 52 -> 413 | Improved |
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.
- Poseidon — POSEidon: Face-From-Depth for Driver Pose Estimation (CVPR 2017) · 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
- ExtPose: Robust and Coherent Pose Estimation by Extending ViTs (ICML 2025 · similarity 0.90)
- SVTformer: Spatial-View-Temporal Transformer for Multi-View 3D Human Pose Estimation (AAAI 2025 · similarity 0.88)
- Hourglass Tokenizer for Efficient Transformer-Based 3D Human Pose Estimation (CVPR 2024 · similarity 0.87)
- Optimizing Human Pose Estimation Through Focused Human and Joint Regions (AAAI 2025 · similarity 0.86)
- SpatioTemporal Learning for Human Pose Estimation in Sparsely-Labeled Videos (AAAI 2025 · similarity 0.86)
- Video-Based Human Pose Regression via Decoupled Space-Time Aggregation (CVPR 2024 · 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.
