DSPO trains real-world super-resolution models with per-object human-preference pairs and text feedback on hallucinated details, and reports higher human-preference win rates than its baselines on DRealSR and RealSR.
Image quality assessment: Unifying structure and texture similarity
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
DSPO: Direct Semantic Preference Optimization for Real-World Image Super-Resolution
DSPO trains real-world super-resolution models with per-object human-preference pairs and text feedback on hallucinated details, and reports higher human-preference win rates than its baselines on DRealSR and RealSR.