CITY-OSM - ETH Zurich

by No License

CITY-OSM - ETH Zurich

# Learning Aerial Image Segmentation From Online MapsThis is the ground truth data generated for the publicationLearning Aerial Image Segmentation From Online MapsPascal Kaiser, Jan Dirk Wegner, Aurélien Lucchi, Martin Jaggi, Thomas Hofmann, Konrad SchindlerIEEE Transactions on Geoscience and Remote Sensing 55 (11), 6054-6068, 2017/11https://doi.org/10.1109/TGRS.2017.2719738Ground truth of Berlin, Chicago, Paris, Potsdam, and Zurich consist of aerial images from Google Mapsand pixel-wise building, road, and background labels from OpenStreetMap. Ground truth of Tokyo consistsof one aerial image from Google Maps and manually generated, pixel-wise building, road, and background labels.Pixel-wise labels are provided as PNG images in RGB order. Pixels labeled as building, road, andbackground are indicated by RGB colors [255,0,0], [0,0,255], and [255,255,255].RGB channel means of aerial imagesBerlin R: 79.94162, G: 84.72064, B: 78.94711Chicago R: 86.46459, G: 85.73488, B: 77.14777Paris R: 82.46727, G: 92.82243, B: 88.05664Potsdam R: 74.85480, G: 77.37761, B: 70.22035Tokyo R: 96.96883, G: 98.44344, B: 108.60135Zurich R: 62.36962, G: 66.11001, B: 60.32863Ground truth was generated inBerlin Spring 2016Chicago Autumn 2015Paris Autumn 2015Potsdam Spring 2016Tokyo Spring 2017Zurich Autumn 2015Ground truth of Potsdam covers the same area as the publicly avaialble, manually labeled ISPRS groundtruthISPRS semantic labeling challengehttp://www2.isprs.org/commissions/comm3/wg4/semantic-labeling.html

Dataset Attributes

Label SVG
CategoriesComputer, Vision, Aerial, Image, Map, Geoscience, Remote, Sensing, Deep, Learning, Berlin, Chicaco, Paris, Potsdam, Tokyo, Zurich