Prompting Multi-Modal Image Segmentation with Semantic Grouping

云想衣裳花想容，春风拂槛露华浓