Multimodal urban areas of interest generation via remote sensing imagery and geographical prior

Chuanji Shi, Yingying Zhang, Jiaotuan Wang, Xin Guo, Qiqi Zhu

Abstract: Highlights•A multimodal model for area-of-interest generation and reliability validation.•A Transformer network aggregates remote sensing imagery and geographical priors.•An enhanced content query in decoder for integrating multimodal information.•A geo-aware regression head for direct prediction of polygonal vertices.