LLM-Optic: Unveiling the Capabilities of Large Language Models for Universal Visual Grounding | Synapse