In conducting damage assessment for scenario earthquakes in high seismic risk regions, building inventory data are required as well as building fragility functions and strong-motion distributions. But inventory data with the locations and characteristics of buildings are not so easy to construct, especially for developing countries. Hence in this study, an approach to construct building inventory data is sought as an alternative of cadastral data and field surveys. Using a high-resolution optical satellite image acquired by WorldView-2, this paper tries to develop building inventory data for earthquake damage assessment in Tacna, Peru. First, Pixel-based classification was carried out to examine basic land-cover and land-use of the urban area. Object-based building extraction was then conducted for three selected areas as an attempt to develop building inventory data.