Text-Guided Few-Shot Semantic Segmentation with Training-Free Multimodal Feature MatchingGuillaume ButhmannTomoya Sakaiet al.2025ICASSP 2025