TopViewRS: Vision-Language Models as Top-View Spatial Reasoners (EMNLP 2024 Oral) - View it on GitHub