Experimental fluvial-deltaic stratigraphic patches for machine learning applications

Collection of 6,132 images (128 x 128 pixels) cropped from experimental stratigraphy produced in the Tulane Delta Basin, TDB-10-1, under temporally constant boundary conditions. The images are prepared to be used in a machine learning project.

Each image is prefixed with a number [0-5] which indicates the strike section the image was selected from. The cropped strike sections are obtained from the archival dataset located on SEN: http://sedexp.net/catalog/tdb-10-1-tulane-delta-basin.

After cropping from the strike sections, each image was processed with binarization and a sequence of morphological opening and closing operations. The code that did the processing can be obtained at https://github.com/amoodie/StratGAN/blob/master/process_images/nrand_process.py.

This data was produced as part of a larger project: https://github.com/amoodie/StratGAN