Multi-Image Super-Resolution for Remote Sensing using Deep RecurrentNetworks

Pytorch implementation of MISR-GRU, a deep neural network for multi image super-resolution (MISR), for ProbaV Super Resolution Competition European Space Agency's Kelvin competition.

MISR-GRU Architecture

*** Trained model is available to download (https://github.com/rarefin/MISR-GRU/blob/master/resources/MISR-GRU.pth)

Example of Super Resolution

A recipe to enhance the vision of the ESA satellite Proba-V

0. Setup python environment

• Setup a python environment and install dependencies, we need python version >= 3.6.8

pip install -r requirements.txt

1. Download data and save clearance

• Download the data from the Kelvin Competition and unzip it

• Run the save_clearance script to pre-compute clearance scores for low-res views

python save_clearance.py --data_dir /path/to/ESA_data

2. Train model

• Train a model with default config

python train.py --config_file_path ../config.json

3. Test model - Create Submission file

• Train a model with default config

python create_submission_file.py --config_file_path ../config.json

3. Submit result and check performance

Although comepetetion is over but model performance PROBA-V Super Resolution post mortem

Authors

Md Rifat Arefin, Samira E. Kahou, Vincent Michalski Alfredo Kalaitzis