OVMR: Open-Vocabulary Recognition with Multi-Modal References | Read Paper on Bytez