.png){kind=icon}
.png){kind=icon}
Return
Task-aware all-in-one guided image super-resolution
.png){kind=icon}
PRE
AI
DOI：10.1016/j.patcog.2026.113483.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• A unified all-in-one GISR framework: We propose the first unified architecture that accommodates a diverse set of GISR subtasks. By bridging the modality and task gaps, MAG-Net offers a generalizable solution for cross-domain guided SR. • A novel task-aware guidance mechanism: Our design incorporates multi-modal prompt generation and adaptive feature routing through MoE, enabling precise, task-specific feature enhancement. • Extensive evaluation with state-of-the-art results: We conduct comprehensive experiments on representative GISR benchmarks, demonstrating that MAG-Net consistently achieves superior performance compared to existing methods.
Keywords:
GISR
MAG-Net
task-aware guidance
image super-resolution
multi-modal prompt
Journal
IF:
7.6
Papers:
1.3W
Citations:
4.5W
Organization
No organization information available