Grasp Any Region: Towards Precise, Contextual Pixel Understanding for Multimodal LLMs

Open in new window