CVIAN: Cross-View Geolocalization and Disaster Mapping with Street-View and VHR Satellite Imagery for Hurrican IAN

CVIAN is a cross-view dataset to support geolocalization and disaster mapping with street-view and very high resolution (VHR) satellite imagery in Florida, USA after Hurricane IAN in 2022. CVIAN contains 4,121 pairs of street-view and VHR satellite imagery, which are manually classified into 3 classes (i.e., light, medium, and heavy damage). The VHR satellite imagery was originally provided by the National Oceanic and Atmospheric Administration (NOAA) at a spatial resolution of 30cm per pixel on September 30, 2022. The street-view imagery was collected from the open-source Mapilliary platform, speicifally from a mapping campaign done by Site Tour 360 after Hurricane IAN hit the study area. CVIAN is the first of this kind dataset that can support both cross-view geolocalization and disaster mapping at the same time.

搜集汇总

背景与挑战

背景概述

CVIAN是一个跨视角数据集，用于支持2022年飓风IAN后美国佛罗里达州的地理定位和灾害测绘，包含4121对街景和极高分辨率卫星图像，并手动分类为三个损害等级。该数据集首次同时支持跨视角地理定位和灾害测绘，数据来源于NOAA的卫星图像和Mapilliary平台的街景图像。