This repo contains the reproduced and adapted work of the paper HighRes-net: [Recursive Fusion for Multi-Frame Super-Resolution of Satellite Imagery].(https://arxiv.org/abs/2002.06460)
- Network: Contains all the source and configuration files needed to run HighRes-Net. These files have been modified from the original version to be adapted to work with the Virtual Moon Dataset.
- Utils: Contains the utility scripts that were created to automate several processes:
- lro_url_download.py: Reads a list of urls from a CSV file and download the files of those links into a specified directory.
- database: Contains all the scripts related to the Virtual Moon database creation:
- 8to16.py: Converts images from 8 bits to 16 bits.
- database.py: Converts into greyscale the images inside every dataset in the directory and includes the clearance maps needed for the
- lro_image_slicer: Crops a selected LRO image into small slices after modifying the intensity 3 times to create a validation dataset of a total of 20 LR images for the network.
- monochannel.py: Converts images from color (3 channels) to greyscale (1 channel).
- slicer.py: Creates datasets containing one HR and 20 LR images per input
Additional information