Return
Multi-conditional Diffusion Action Generation Network for Human Action Recognition
DOI:10.1016/j.inffus.2026.104243.png)
Abstract
En 中文
• MD-AGN is proposed to synthesize human actions with a few samples. • HPRE constructs a multi-conditional prompting guide from text to action. • MMDM employs anisotropy factors to learn irregular distributions in the action. • IBO optimizes classification boundaries by excluding bad samples.
Keywords:
MD-AGN
HPRE
MMDM
IBO
Journal
IF:
15.5
Papers:
4.2K
Citations:
2.7W
Organization
Cited Papers
SRAGAN: Saliency regularized and attended generative adversarial network for Chinese ink-wash painting style transfer
PATTERN RECOGNITION
IF7.6

