jain15mayank / GAN-augmentation-cloud-image-segmentation

Using GANs to Augment Data for Cloud Image Segmentation Task

With the spirit of reproducible research, this repository contains all the codes required to produce the results in the manuscript:

Jain, M., Meegan, C. and Dev, S.(2021). Using GANs to Augment Data for Cloud Image Segmentation Task. In: IEEE International Geoscience and Remote Sensing Symposium (IGARSS), 2021

As explained in the manuscript, the SWINSEG dataset is used to carry out the experiments. The description of each of the code files is as follows:

• GAN.py: Reads the sky/cloud images, trains a GAN and then use it to generate the new images
• clustering.py: Contains the code to perform sky/cloud image segmentation using k-Means clustering (unsupervised)
• transformations.py: Contains some utility functions to perform basic image transformations for data augmentation
• smoothBinMaps.py: Smoothens the segmentation maps that were estimated using the clustering.py
• main.py: Trains and evaluate PLS regression method with and without GAN augmentation. It further checks if the generated sky/cloud & GT map pair falls within the distribution of the original dataset.