I am trying to apply super resolution models to MODIS 500m imagery in order to downscale it’s resolution down to Sentinel-2’s 60m spectral bands.
I know it is a very challenging task to do, since I have a dataset consisting only of 20000 images and I haven’t found out so far anything like that in the literature.
I have tried implementing several architectures, from simple CNNs to more complicated SRGANs, but my results are far from the expected outputs. Do you have any suggestions?
New contributor
dry_martini is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.