Scholarlink

<p>Human action recognition is pivotal in computer vision, with applications ranging from surveillance to human–robot interaction. Despite the effectiveness of supervised skeleton-based methods, their reliance on exhaustive annotation limits generalization to novel actions. Zero-shot skeleton action recognition emerges as a promising paradigm, yet it faces challenges due to the spectral bias of diffusion models, which oversmooth high-frequency dynamics. Here, we propose frequency-aware diffusion for skeleton-text matching, integrating a semantic-guided spectral residual module, a timestep-adaptive spectral loss, and curriculum-based semantic abstraction to address these challenges. Our approach effectively recovers fine-grained motion details, achieving state-of-the-art performance on NTU RGB+D, PKU-MMD, and Kinetics-skeleton datasets. Code has been made available at <a href="https://github.com/yuzhi535/FDSM">https://github.com/yuzhi535/FDSM</a>. Project homepage: <a href="https://yuzhi535.github.io/FDSM.github.io/">https://yuzhi535.github.io/FDSM.github.io/</a>.</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

Frequency-enhanced diffusion models: curriculum-guided semantic alignment for zero-shot skeleton action recognition

  • 期刊论文

  • 发表日期：

    2026年5月22日

查看全文
  • Yuxi Zhou,
  • Zhengbo Zhang,
  • Jingyu Pan,
  • Zhiyu Lin,
  • Zhigang Tu

摘要

Human action recognition is pivotal in computer vision, with applications ranging from surveillance to human–robot interaction. Despite the effectiveness of supervised skeleton-based methods, their reliance on exhaustive annotation limits generalization to novel actions. Zero-shot skeleton action recognition emerges as a promising paradigm, yet it faces challenges due to the spectral bias of diffusion models, which oversmooth high-frequency dynamics. Here, we propose frequency-aware diffusion for skeleton-text matching, integrating a semantic-guided spectral residual module, a timestep-adaptive spectral loss, and curriculum-based semantic abstraction to address these challenges. Our approach effectively recovers fine-grained motion details, achieving state-of-the-art performance on NTU RGB+D, PKU-MMD, and Kinetics-skeleton datasets. Code has been made available at https://github.com/yuzhi535/FDSM. Project homepage: https://yuzhi535.github.io/FDSM.github.io/.