Training temporal Convolution Neural Netoworks (TempCNNs), Recurrent Neural Networks (RNNs) and Random Forests (RFs) on satelitte image time series. This code is supporting a paper submitted to IEEE International Geoscience and Remote Sensing Symposium (IGARSS) 2019:
@inproceedings{Pelletier2018Deep,
Title = {Deep Learning for the classification of Sentinel-2 image time series},
Author = {Pelletier, Charlotte and Webb, Geoffrey I and Petitjean, Francois},
Booktitle = {IEEE International Geoscience and Remote Sensing Symposium (IGARSS) 2019},
note = {Accepted for an oral presentation}
}
- training TempCNNs:
python train_classifier.py --classifier TempCNN --train train_dataset.csv --test test_dataset.csv - training bidirectional GRU-RNNs:
python train_classifier.py --classifier GRU-RNNbi --train train_dataset.csv --test test_dataset.csv - training GRU-RNNs:
python train_classifier.py --classifier GRU-RNN --train train_dataset.csv --test test_dataset.csv - training RFs:
python train_classifier.py --classifier RF --train train_dataset.csv --test test_dataset.csv
It will output a result file including the OA computed on test data, the confusion matrix, the training history for deep learning models, and the learned model.
Each model will be trained on train_dataset.csv file and test on test_dataset.csv file.
Please note that both train_dataset.csv and test_dataset.csv files are a subsample of the data used in the paper: original data cannot be distributed.
Thoses files have an header, and contain one observation per row having the following format:
[class,objectID,date1.B2,date1.B3,date1.B4,date1.B5,date1.B6,date1.B7,date1.B8,date1.B8A,date1.B11,date1.B12,...,date73.B12]
The produced map for TempCNNs, bidirectional GRU-RNNs, and RFs are available in the map folder.
You can also have a look to our full map of Victoria (Australia) here.