New benchmark and model help AI compare urban images and text

Original: Geo-Embed: Towards Unified Multimodal Embeddings for Urban Understanding

Source: arxiv.org ↗