Frequency-Enhanced Diffusion Models: Curriculum-Guided Semantic Alignment for Zero-Shot Skeleton Action Recognition
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913177437995008 |
|---|---|
| author | Zhou, Yuxi Zhang, Zhengbo Pan, Jingyu Lin, Zhiyu Tu, Zhigang |
| author_facet | Zhou, Yuxi Zhang, Zhengbo Pan, Jingyu Lin, Zhiyu Tu, Zhigang |
| contents | 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 (ZSAR) 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 (FDSM), 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/ |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_09063 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Frequency-Enhanced Diffusion Models: Curriculum-Guided Semantic Alignment for Zero-Shot Skeleton Action Recognition Zhou, Yuxi Zhang, Zhengbo Pan, Jingyu Lin, Zhiyu Tu, Zhigang Computer Vision and Pattern Recognition Artificial Intelligence 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 (ZSAR) 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 (FDSM), 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/ |
| title | Frequency-Enhanced Diffusion Models: Curriculum-Guided Semantic Alignment for Zero-Shot Skeleton Action Recognition |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2604.09063 |