Data for the ResGNN paper

This repository contains data for the ResGNN paper as follows: 1) raw_volumes.zip - binary 200^3-voxel sub-volumes in the .raw format. In the same folder, README describes the order of rock types, while labels.npy contains porosity and permeability values for each sub-volume. 2) individual_pkl - pre-calculated graphs for each rock type separately. We provide it to give potential users the flexibility to form their own train/unseen splits. 3) train_unseen_pkl_datasets.zip - pre-calculated graphs for training and unseen datasets used in the paper.