Because very often, in the literature, manually generation of OSM data is described as a complex and questioned process, it might be useful to support it by automated remote sensing approaches. While automated approaches are not expected to deliver fully accurate results, they still could save important work which may substantially reduce the digitizing efforts. The aim of this session is to present our developed building extraction algorithm that combines an object-based (OBIA) approach with a deep learning algorithm: convolutional neural network (CNN). The results (building polygons) from the OBIA algorithm are used as training samples for CNN algorithm. The buildings extraction algorithm will be further tested for usefulness in supporting the OSM initiative, particularly in areas exposed to humanitarian crisis where OSM data does not exist at all. The algorithm has been tested on RGB images. In all cases, it has provided promising results, with an overall accuracy over 85%.