LocateAnything: Fast Vision-Language Grounding with Parallel Box Decodingresearch.nvidia.com2 pointsgmays11 days ago