SalArt-VQA benchmark shows that high image-level artifact detection accuracy in VLMs does not imply correct localization, grounding, or evidence-supported defect descriptions.
arXiv preprint arXiv:2602.09475 , year=
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
-
SalArt-VQA: Diagnosing Whether VLMs Understand Salient Artifacts in Generated Images
SalArt-VQA benchmark shows that high image-level artifact detection accuracy in VLMs does not imply correct localization, grounding, or evidence-supported defect descriptions.