2026-10-06T12:04:04Z https://api.figshare.com/v2/oai

oai:figshare.com:article/34017653 2026-09-28T17:52:42Z category_1 category_16 category_21 category_272 category_734 portal_5 item_type_3 month_year_09_2026

<p>Hierarchical joint error (MPJPE & P-MPJPE) comparison on Human3.6M.</p> Jinze Jiang (25132820) Zhongheng Jian (25132823) Hualing Zheng (25132826) Minsi Liang (23393861) Yuxing Shi (10150023) Liming Zhang (213971) Biophysics Physiology Biotechnology Environmental Sciences not elsewhere classified Biological Sciences not elsewhere classified two features separately skeletal kinematic tree realize dual supervision injects gaussian noise increased estimation error grained motion analysis athletic technique diagnosis applications including vr alleviate error amplification sports motion capture lightweight inference configurations 3d pose loss mainstream methods fail physical fitness assessment outperforms mainstream diffusion full diffusion pipeline based probabilistic methods based methods suffer digital human animation sports rehabilitation training monocular 3d hpe hypothesis inference setting reduces average mpjpe reaching new state forward diffusion process bone direction features 3dhp datasets show reverse denoising process grained hierarchical spatio based methods sports training hypothesis inference 3d hpe diffusion process art methods quantitative assessment hierarchical spatial diffusion model temporal denoising immersive human human skeleton hierarchical spatio work provides temporal relationships temporal correlations joint levels fully mine <p>Hierarchical joint error (MPJPE & P-MPJPE) comparison on Human3.6M.</p> 2026-09-28T17:52:13Z Dataset Dataset 10.1371/journal.pone.0359321.t008 https://figshare.com/articles/dataset/_p_Hierarchical_joint_error_MPJPE_P-MPJPE_comparison_on_Human3_6M_p_/34017653 CC BY 4.0