As mentioned in the problem description, the images in the test set have been generated without valid georeferencing. “The correct georeferences for the test chips have been removed”. I am doing some research on the influence of remote sensing data scale on deep learning models and am interested in the spatial/geodetic resolution of the test images. Would it be possible to get the original images the test set was tiled from with correct georeferences?
Hi @Damjanh - Have you checked the data on Radiant MLHub? You may find it helpful to follow up there. See here for more info Competition: Open Cities AI Challenge: Segmenting Buildings for Disaster Resilience