RA-L · 2024FM-FusionAbstract[RA-L] FM-Fusion: Instance-aware Semantic Mapping Boosted by Vision-Language Foundation ModelsRepositoryarXiv