Semi-supervised Representation Learning for Remote Sensing Image Classification Based on Generative Adversarial Networks

Nov 18, 2021 1 min read

SSRL-for-image-classification

The installed packages are listed is conda.yaml.

The whole training and test procedure is as follows.

creat a conda env and install all necessary packages.
run data_split.py to create five folders of dataset under five-folder cross verification protocol.
run mixM2_Res.py to train model and save models.
run extract_feature.py to extract features from images.
train_svm.py to train svm and test.
Utils.py implement the data augmentation. Network.py defines the network architecture of Generator and Discriminator.

This code borrow heavily from MATRA GANs. If you find this repository useful, please cite the following paper.

Yan Peiyao; He Feng*; Yang Yajie; Hu Fei; Semi-Supervised Representation Learning for Remote Sensing Image Classification Based on Generative Adversarial Networks, IEEE Access, 2020, 8: 54135-54144.

GitHub

View Github

Network Remote

John was the first writer to have joined pythonawesome.com. He has since then inculcated very effective writing and reviewing culture at pythonawesome which rivals have found impossible to imitate.