EDDA - Mozambique post disaster building damage dataset

Abstract

Natural disasters have devastating effects on communities, necessitating swift and accurate damage assessment. Manual assessment methods are time-consuming and costly, emphasizing the significance of semiautomatic ap- proaches employing remote sensing and drone data. However, current datasets primarily focus on Western countries’ infrastructure, lacking information on damaged buildings in other regions specifically Africa. To bridge this gap, we present the EDDA dataset, comprising of VHR orthorectified mosaic images from drone imagery with building footprint labels classified by damage extent of rural and urban areas of Mozambique affected by Cyclone Ida.