High-resolution satellite imagery is changing our understanding of the world around us, as well as the way we as humans interact with our planet. However, the raw images do little more than pique our interest — unless we can superimpose a layer that actually identifies real objects.
Reliable labeling of buildings based on satellite imagery is one of the first and most challenging steps in producing accurate 3D models and maps. While automated algorithms continue to improve, significant manual effort is still necessary to ensure geospatial accuracy and acceptable quality. Improved automation is required to enable more rapid response to major world events such as humanitarian and disaster response. 3D height data can help improve automated building labeling performance, and capabilities for providing this data on a global scale are now emerging. In this challenge, we ask solvers to use satellite imagery and newly available 3D height data products to improve upon the state of the art for automated building detection and labeling.
USSOCOM is seeking an algorithm that provides reliable, automatic labeling of buildings based solely on orthorectified color satellite imagery and 3D height data.
What data will be provided to you?
Competitors will receive an orthorectified color image, Digital Surface Model (DSM), and Digital Terrain Model (DTM) for each geographic area of interest (AOI). The DSM indicates the height of the earth, with objects such as buildings and trees included. The DTM indicates only the height of the ground. Both should be expected to include some errors, and errors may be expected to be similar in the provisional and sequestered data sets. The difference in the DSM and DTM indicates height of objects above ground. All input files provided are raster GeoTIFF images. Ground truth building labels will also be provided for a subset of the data to be used for training.
Prizes and Conditions
The top five solutions on the provisional leaderboard will be asked to submit their software for independent evaluation with sequestered test data to establish the final leaderboard for prize award.
Software must be completely automated and rely only on the input color ortho image, DSM, and DTM provided for each scene.
Machine learning models are acceptable (and encouraged) as an input as long as the models are delivered such that the submitted software can be successfully executed on sequestered test data. Software may rely on open source third-party libraries as long as all necessary dependencies are delivered with documentation such that the software can be successfully built and executed to confirm that it produces the submitted solution.