{"ID":2876628,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2508.21363","arxiv_id":"2508.21363","title":"Efficient Diffusion-Based 3D Human Pose Estimation with Hierarchical Temporal Pruning","abstract":"Diffusion models have demonstrated strong capabilities in generating high-fidelity 3D human poses, yet their iterative nature and multi-hypothesis requirements incur substantial computational cost. In this paper, we propose an Efficient Diffusion-Based 3D Human Pose Estimation framework with a Hierarchical Temporal Pruning (HTP) strategy, which dynamically prunes redundant pose tokens across both frame and semantic levels while preserving critical motion dynamics. HTP operates in a staged, top-down manner: (1) Temporal Correlation-Enhanced Pruning (TCEP) identifies essential frames by analyzing inter-frame motion correlations through adaptive temporal graph construction; (2) Sparse-Focused Temporal MHSA (SFT MHSA) leverages the resulting frame-level sparsity to reduce attention computation, focusing on motion-relevant tokens; and (3) Mask-Guided Pose Token Pruner (MGPTP) performs fine-grained semantic pruning via clustering, retaining only the most informative pose tokens. Experiments on Human3.6M and MPI-INF-3DHP show that HTP reduces training MACs by 38.5\\%, inference MACs by 56.8\\%, and improves inference speed by an average of 81.1\\% compared to prior diffusion-based methods, while achieving state-of-the-art performance.","short_abstract":"Diffusion models have demonstrated strong capabilities in generating high-fidelity 3D human poses, yet their iterative nature and multi-hypothesis requirements incur substantial computational cost. In this paper, we propose an Efficient Diffusion-Based 3D Human Pose Estimation framework with a Hierarchical Temporal Pru...","url_abs":"https://arxiv.org/abs/2508.21363","url_pdf":"https://arxiv.org/pdf/2508.21363v3","authors":"[\"Yuquan Bi\",\"Hongsong Wang\",\"Xinli Shi\",\"Zhipeng Gui\",\"Jie Gui\",\"Yuan Yan Tang\"]","published":"2025-08-29T07:08:07Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[\"Diffusion Model\"]","has_code":false}