Fetching the paper…

Learning to Generate Text-grounded Mask for Open-world Semantic Segmentation from Only Image-Text Pairs · Around