A CNN-Based Fusion Method for Feature Extraction from Sentinel Data
A CNN is trained to perform the estimation of the NDVI, using coupled Sentinel-1 and Sentinel-2 time-series.
- Giuseppe Scarpa;
- Massimiliano Gargiulo;
- Antonio Mazza, contact person;
- Raffaele Gaetano.
Copyright (c) 2018 Image Processing Research Group of University Federico II of Naples, 'GRIP-UNINA'.
All rights reserved. This work should only be used for nonprofit purposes.
By downloading and/or using any of these files, you implicitly agree to all the terms of the license, as specified in the document LICENSE.txt (included in this directory).
This code is written for Python2.7 and uses Theano and Lasagne libraries. The list of all requirements is in requirements.txt.
The command to install the requirements is:
cat requirements.txt | xargs -n 1 -L 1 pip2 install
Optional requirements for using gpu:
- cuda = 8
- cudnn = 5
Firstly, you have to download the dataset from LINK.
The 8 proposed techniques to estimate the NDVI are:
SAR, SARp, OPTI, OPTII, SOPTI, SOPTII, SOPTIp, SOPTIIp. (1)
In the paper, these techniques correspond respectively to:
SAR, SAR+, Optical/C, Optical, Optical-SAR/C, Optical-SAR, Optical-SAR+/C, Optical-SAR+.
To train and/or test the CNN you have to use the Train.py and/or Test.py, in TRAINING and TEST directory respectively.
In these files you have to set the technique_name that you can choose from (1):
kwargs['identifier'] = 'technique_name'
If you use this CNN-based approach in your research or wish to refer to the baseline results, please use the following BibTeX entry.
@article{scarpa2018cnn,
title={A CNN-Based Fusion Method for Feature Extraction from Sentinel Data},
author={Scarpa, Giuseppe and Gargiulo, Massimiliano and Mazza, Antonio and Gaetano, Raffaele},
journal={Remote Sensing},
volume={10},
number={2},
pages={236},
year={2018},
publisher={Multidisciplinary Digital Publishing Institute}
}