Current image-text alignment metrics, including CLIPScore and DSGScore, produce unstable model rankings under random seeds and are highly sensitive to tiny image perturbations.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Re-Thinking the Automatic Evaluation of Image-Text Alignment in Text-to-Image Models
Current image-text alignment metrics, including CLIPScore and DSGScore, produce unstable model rankings under random seeds and are highly sensitive to tiny image perturbations.