Zero-shot Quality Representation Learning (EyeQC)

Learning image-quality representations from frozen vision-language embeddings