The NTIRE 2026 RAIM challenge introduces a benchmark where MLLMs must select the superior image from high-quality pairs and provide grounded expert-level explanations for the choice.
iDiff: Interpretable Difference-aware Framework for Pairwise Image Quality Assessment
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
NTIRE 2026 The 3rd Restore Any Image Model (RAIM) Challenge: Professional Image Quality Assessment (Track 1)
The NTIRE 2026 RAIM challenge introduces a benchmark where MLLMs must select the superior image from high-quality pairs and provide grounded expert-level explanations for the choice.