Structural attributes derived from Google Street View imagery, Louisiana coastal zone

<p>When developing plans for flood risk mitigation, one of the first hurdles local and state regional decision-makers must clear is gathering data on what assets actually exist on the ground in their jurisdictions. Even in areas like Louisiana—where significant effort has been taken to study flood risk since Hurricane Katrina struck in 2005—property data sets are incomplete, fragmented across agencies and jurisdictions, and obsolete. This data set, funded by the Andrew W Mellon Foundation, uses automated image processing of Google Street View pictures to extract building attributes relevant to flood risk, such as the foundation height and type, square footage of the building footprint, number of stories, and usage (e.g., residential, commercial). The classification and inference of these attributes were produced using machine learning methods for the purpose of developing structure-level estimates of risk that have been incorporated into a data viewer and decision support system. Risk estimates are derived from these building characteristics and hazard estimates from Louisiana's 2017 Coastal Master Plan. </p>