Omni-Q: Omni-Directional Scene Understanding for Unsupervised Visual Grounding | Read Paper on Bytez