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

VLM
Re-assign community