Scale-MAE: A Scale-Aware Masked Autoencoder for Multiscale Geospatial Representation Learning | IEEE Conference Publication | IEEE Xplore