Return
Strengthening temporal action segmentation through diffusion models
DOI:10.1016/j.engappai.2025.113163.png)
Abstract
En 中文
• Our enhanced frame-level feature extraction further improves the action segmentation task. • A novel diffusion-based approach improves the feature robustness and diversity. • The multi-stage and multi-task strategies achieve targeted fine-tuning of frame-level feature extraction.
Journal
IF:
8
Papers:
5.3K
Citations:
3.5W

