The Christchurch Aerial Semantic Dataset (CASD) comprises aerial imagery at very high resolution over Christchurch, New-Zealand, and reference semantic data for urban objects such as buildings, vegetation and vehicles. It aims to foster developments of new methods for Earth observation including automatic image classification, object detection, semantic segmentation, semi-supervised learning, etc... Aerial imagery consists in the Christchurch Earthquake Imagery dataset, released by Land Information New Zealand: https://www.linz.govt.nz/land/maps/linz-topographic-maps/imagery-orthophotos/christchurch-earthquake-imagery Annotations were produced by ONERA/DTIS on 4 images. Three classes were tagged: buildings (797 objects), cars (2357 objects), and vegetation (938 objects). All objects are given as polygonal bounding boxes (shapefiles) and image semantic masks (rasters). License This imagery is licensed under a Creative Commons Attribution 4.0 International licence (https://creativecommons.org/licenses/by/4.0/) with Crown copyright reserved. This means anyone is free to copy, distribute, and adapt the imagery so long as it is attributed to the Crown, eg ”Crown Copyright Reserved.” The annotations are licensed also under a Creative Commons Attribution 4.0 International licence (https://creativecommons.org/licenses/by/4.0/) with authors’ and ONERA copyright reserved. This means anyone is free to copy, distribute, and adapt the annotations as long as they are attributed to the authors and ONERA. For research papers, acknowledgments can be done by citing the authors’ works: H. Randrianarivo, B. Le Saux, and M. Ferecatu. Man-made structure detection with deformable part-based models. In IEEE Int. Geoscience and Remote Sensing Symposium (IGARSS), Melbourne, Australia, July 2013 N. Audebert, B. Le Saux, and S. Lefèvre. Segment-before-Detect: Vehicle Detection and Classification through Semantic Segmentation of Aerial Images. Remote Sensing, 9(4):1–18, April 2017

{"references": ["Randrianarivo, H. et al. (2013), Man-made structure detection with deformable part-based models, Proc. IGARSS", "Randrianarivo, H. et al. (2015). Discriminatively- trained model mixture for object detection in aerial images. Proc. IIM", "Audebert et al. (2017). Segment-before-Detect: Vehicle Detec- tion and Classification through Semantic Segmentation of Aerial Images, Remote Sensing, 9(4)"]}