A massive data store of 3D geological models for Machine Learning & inversion applications

There is always a problem finding a good dataset for training a neural network. Open field datasets are often too restricted for one particular use (facies, faults, etc.) as well as open synthetic datasets are of mediocre quality and lack variability.

I've stumbled upon a paper that explains how the authors created a massive dataset of 3D geological models for Machine Learning and Inversion applications using the Noddy modeling platform. In the comments, you can find a link for the Cloudstor page, where you can find 1M models (200x200x200 voxels per model) and the history file used to define each model.

Found in SWUNG slack channel.

Jessell, M., Guo, J., Li, Y., Lindsay, M., Scalzo, R., Giraud, J., Pirot, G., Cripps, E., and Ogarko, V.: Into the Noddyverse: A massive data store of 3D geological models for Machine Learning & inversion applications, Earth Syst. Sci. Data Discuss. [preprint], https://lnkd.in/dCfkpYJp, in review, 2021.

Link to the dataset: https://cloudstor.aarnet.edu.au/plus/s/8ZT6tjOvoLWmLPx
Link to the history files: https://cloudstor.aarnet.edu.au/plus/s/UxnVSkHfnr7chW9
You can open history files using this: https://tectonique.net/noddy/
Link to the paper: https://essd.copernicus.org/preprints/essd-2021-304/
Link to the LinkedIn post