When Visual Grounding Meets Gigapixel-level Large-scale Scenes: Benchmark and Approach | Read Paper on Bytez