1 paper
Xinyue Xu, Zheng Zhang, Kunyang Ma +5
The paper introduces DM-KG, a direction‑metric knowledge graph that extracts 3D spatial relationships from street‑view images and injects them into vision‑language models to improv…