An MLLM-powered visual-semantic indicator system coupled with geographically weighted XGBoost for precise, spatially explicit urban fire-risk governance in Wuhan.
A multi-perspective NDVI-GVI framework revealing how urban expansion reshapes the spatial relationships between urban green space and urban morphology in Guangzhou.
We proposed GeoIdentity-Sat2Street, a geographic identity preserving satellite-to-street-view generation framework that expands street view coverage in data-scarce regions.
We develop street view imagery-based urban visual intelligence, from SVI coverage generation to GeoAI and LLM visual-semantic sensing, to quantify eye-level built environments and support urban management.