Tag
CIPER is a unified transformer framework that jointly performs city-scale retrieval and precise 3-DoF pose estimation from cross-view images, overcoming limitations of cascade pipelines.