# Beyond Static Frames: Temporal Aggregate-and-Restore Vision Transformer for Human Pose Estimation

arXiv: https://arxiv.org/abs/2603.05929 · HTML: https://mlsanity.com/paper/2603.05929/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 38 of its 44 references that are in the map.
- First posted on arXiv 6 Mar 2026.

## Builds on (most cited first, 30 of 38)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks](https://mlsanity.com/paper/1506.01497/) (2015)
- [Masked-attention Mask Transformer for Universal Image Segmentation](https://mlsanity.com/paper/2112.01527/) (2021)
- [Deep High-Resolution Representation Learning for Human Pose Estimation](https://mlsanity.com/paper/1902.09212/) (2019)
- [Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields](https://mlsanity.com/paper/1611.08050/) (2016)
- [Simple Baselines for Human Pose Estimation and Tracking](https://mlsanity.com/paper/1804.06208/) (2018)
- [Convolutional Pose Machines](https://mlsanity.com/paper/1602.00134/) (2016)
- [Sapiens: Foundation for Human Vision Models](https://mlsanity.com/paper/2408.12569/) (2024)
- [PoseTrack: A Benchmark for Human Pose Estimation and Tracking](https://mlsanity.com/paper/1710.10000/) (2017)
- [Human Pose Regression with Residual Log-likelihood Estimation](https://mlsanity.com/paper/2107.11291/) (2021)
- [Joint Training of a Convolutional Network and a Graphical Model for Human Pose Estimation](https://mlsanity.com/paper/1406.2984/) (2014)
- [TransPose: Keypoint Localization via Transformer](https://mlsanity.com/paper/2012.14214/) (2020)
- [Detailed 2D-3D Joint Representation for Human-Object Interaction](https://mlsanity.com/paper/2004.08154/) (2020)
- [Pose Flow: Efficient Online Pose Tracking](https://mlsanity.com/paper/1802.00977/) (2018)
- [Detect-and-Track: Efficient Pose Estimation in Videos](https://mlsanity.com/paper/1712.09184/) (2017)
- [Deep Dual Consecutive Network for Human Pose Estimation](https://mlsanity.com/paper/2103.07254/) (2021)
- [4D Association Graph for Realtime Multi-person Motion Capture Using Multiple Video Cameras](https://mlsanity.com/paper/2002.12625/) (2020)
- [Combining detection and tracking for human pose estimation in videos](https://mlsanity.com/paper/2003.13743/) (2020)
- [Human Pose as Compositional Tokens](https://mlsanity.com/paper/2303.11638/) (2023)
- [PoseTrack: Joint Multi-Person Pose Estimation and Tracking](https://mlsanity.com/paper/1611.07727/) (2016)
- [Learning Temporal Pose Estimation from Sparsely-Labeled Videos](https://mlsanity.com/paper/1906.04016/) (2019)
- [Multi-person Articulated Tracking with Spatial and Temporal Embeddings](https://mlsanity.com/paper/1903.09214/) (2019)
- [Learning Dynamics via Graph Neural Networks for Human Pose Estimation and Tracking](https://mlsanity.com/paper/2106.03772/) (2021)
- [DiffPose: SpatioTemporal Diffusion Model for Video-Based Human Pose Estimation](https://mlsanity.com/paper/2307.16687/) (2023)
- [Hierarchical Contrast for Unsupervised Skeleton-based Action Representation Learning](https://mlsanity.com/paper/2212.02082/) (2022)
- [Temporal Feature Alignment and Mutual Information Maximization for Video-Based Human Pose Estimation](https://mlsanity.com/paper/2203.15227/) (2022)
- [Mutual Information-Based Temporal Difference Learning for Human Pose Estimation in Video](https://mlsanity.com/paper/2303.08475/) (2023)

## Abstract (arXiv)

Vision Transformers (ViTs) have recently achieved state-of-the-art performance in 2D human pose estimation due to their strong global modeling capability. However, existing ViT-based pose estimators are designed for static images and process each frame independently, thereby ignoring the temporal coherence that exists in video sequences. This limitation often results in unstable predictions, especially in challenging scenes involving motion blur, occlusion, or defocus. In this paper, we propose TAR-ViTPose, a novel Temporal Aggregate-and-Restore Vision Transformer tailored for video-based 2D human pose estimation. TAR-ViTPose enhances static ViT representations by aggregating temporal cues across frames in a plug-and-play manner, leading to more robust and accurate pose estimation. To effectively aggregate joint-specific features that are temporally aligned across frames, we introduce a joint-centric temporal aggregation (JTA) that assigns each joint a learnable query token to selectively attend to its corresponding regions from neighboring frames. Furthermore, we develop a global restoring attention (GRA) to restore the aggregated temporal features back into the token sequence of the current frame, enriching its pose representation while fully preserving global context for precise keypoint localization. Extensive experiments demonstrate that TAR-ViTPose substantially improves upon the single-frame baseline ViTPose, achieving a +2.3 mAP gain on the PoseTrack2017 benchmark. Moreover, our approach outperforms existing state-of-the-art video-based methods, while also achieving a noticeably higher runtime frame rate in real-world applications. Project page: https://github.com/zgspose/TARViTPose.