Segmenting and Classifying building roofs in high resolution satellite imagery in Bulawayo, Zimbabwe

To monitor the success of the UN Sustainable Development Goals requires the availability of temporally frequent data that can be used to understand the an areas dynamics. Rooftops can be detected in high resolution satellite data and can provide insights into the local socioeconomic situation and monitor its progress over time. This study developed a workflow to classify rooftops in Bulawayo, Zimbabwe with accuracy assessments indicating an accuracy of 85\%. While showing promising results, the classifier showed some difficulty in accurately identifying some smaller classes. With further refinement to the classes and confirmed identification of materials, this method has shown potential for being applied to monitoring development across the city