Scholarlink

<p>Monocular 3D human pose estimation is challenged by depth ambiguity and complex articulation, which complicate feature modeling and demand robust spatio-temporal representations. Although existing methods have advanced spatio-temporal modeling, limitations remain: graph convolutional network (GCN) exhibits low-pass behavior that, as depth increases, attenuates high-frequency geometric details in joint trajectories and thus degrades depth accuracy; and standard self-attention does not explicitly encode skeletal topology, resulting in indirect modeling of bone connectivity. To address these issues, we propose SBAHGNet, a dual-branch spatio-temporal feature-fusion network. In the GCN branch, a Multi-Scale High-Frequency Enhancement (MSHFE) module—applied after feature aggregation-recovers high-frequency geometric cues lost to GCN smoothing, improving fine-grained depth representation. In the attention branch, a Skeletal-Biased Attention (SBA) module injects a learnable skeletal bias into spatial attention to explicitly encode skeletal topology and strengthen structural modeling. Complementary features from both branches are adaptively fused for final 3D pose regression. Extensive experiments on Human3.6&#xa0;M and MPI-INF-3DHP validate our approach. With detected 2D keypoints, SBAHGNet attains 37.24&#xa0;mm MPJPE (P1) and 31.57&#xa0;mm PA-MPJPE (P2) on Human3.6&#xa0;M (12.38&#xa0;mm with ground-truth 2D), and 13.83&#xa0;mm MPJPE, 99.02% PCK@150&#xa0;mm, and 88.22 AUC on MPI-INF-3DHP. With only 18.3&#xa0;M parameters, the model achieves a favorable accuracy–efficiency trade-off and outperforms many comparable methods.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

SBAHGNet:3D human pose estimation via skeleton-biased attention and high-frequency enhanced graph convolution

  • 期刊论文

  • 发表日期：

    2026年6月2日

查看全文
  • Yu Wang,
  • Jiaqiu Ai,
  • Xinyu Sun,
  • Yong Zhang,
  • Jinyang Huang

摘要

Monocular 3D human pose estimation is challenged by depth ambiguity and complex articulation, which complicate feature modeling and demand robust spatio-temporal representations. Although existing methods have advanced spatio-temporal modeling, limitations remain: graph convolutional network (GCN) exhibits low-pass behavior that, as depth increases, attenuates high-frequency geometric details in joint trajectories and thus degrades depth accuracy; and standard self-attention does not explicitly encode skeletal topology, resulting in indirect modeling of bone connectivity. To address these issues, we propose SBAHGNet, a dual-branch spatio-temporal feature-fusion network. In the GCN branch, a Multi-Scale High-Frequency Enhancement (MSHFE) module—applied after feature aggregation-recovers high-frequency geometric cues lost to GCN smoothing, improving fine-grained depth representation. In the attention branch, a Skeletal-Biased Attention (SBA) module injects a learnable skeletal bias into spatial attention to explicitly encode skeletal topology and strengthen structural modeling. Complementary features from both branches are adaptively fused for final 3D pose regression. Extensive experiments on Human3.6 M and MPI-INF-3DHP validate our approach. With detected 2D keypoints, SBAHGNet attains 37.24 mm MPJPE (P1) and 31.57 mm PA-MPJPE (P2) on Human3.6 M (12.38 mm with ground-truth 2D), and 13.83 mm MPJPE, 99.02% PCK@150 mm, and 88.22 AUC on MPI-INF-3DHP. With only 18.3 M parameters, the model achieves a favorable accuracy–efficiency trade-off and outperforms many comparable methods.