Cross-Modal Retrieval Tasks | State-of-the-Art