CLIP (Contrastive Language-Image Pre-Training) is a neural network trained on a variety of (image, text) pairs. It can map images and text into the same latent space, so that they can be compared ...

This paper provides a comprehensive overview of recent advancements in image quality assessment for text-to-image generation. In particular, we review existing quality assessment studies from ...