.png){kind=icon}
.png){kind=icon}
Return
TuckerDreamer: Subject-driven text-to-image generation via Tucker-based frequency-domain fine-tuning
.png){kind=icon}
PRE
AI
Y
B
DOI：10.1016/j.imavis.2026.106088.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• A Tucker decomposition-based method improves subject-driven image generation • Preserving spatial weight structure enhances subject feature extraction • Energy-based masking reduces noise from pre-trained image models • Dynamic adaptive gating mitigates overfitting with limited reference images
Journal
IF:
4.2
Papers:
4.0K
Citations:
6.7K
Organization
No organization information available
Cited Papers