SCENIR: Visual Semantic Clarity through Unsupervised Scene Graph Retrieval | Read Paper on Bytez