Exploiting Contextual Objects and Relations for 3D Visual Grounding | Read Paper on Bytez