Fetching the paper…

Ref-Diff: Zero-shot Referring Image Segmentation with Generative Models · Around