Replication Data for: An object-based approach to differentiate pores and microfractures in petrographic analysis using explainable, supervised machine learning

This dataset contains all the image data used for the study "An object-based approach to differentiate pores and microfractures in petrographic analysis using explainable, supervised machine learning". We also provide the labelled dataset as a CSV file for reference. The dataset of 18 petrographic thin section scans is provided in this repository and also contains no restrictions on its usage. The folder for each of the scans contains the following images; 1. The raw scan 2. The cropped raw scan 3. The processed scan 4. The HSB colorspace thresholded image 5. The HSB colorspace thresholded image with a 30 px closing operation 6. The HSB colorspace thresholded image with a 1000 px opening operation 7. The LAB colorspace thresholded image 8. The LAB colorspace thresholded image with a 30 px closing operation 9. The LAB colorspace thresholded image with a 1000 px opening operation 10. The combined HSB and LAB image (binary) 11. The combined HSB and LAB image (as a color composite) 12. The artifact binary mask (not present in some samples as they contained no visible artifacts) 13. The overall label map 14. The sampling label map (some samples contain multiple versions due to itinerant sampling runs as mentioned in the study) The processing of the images was performed in Fiji (Schindelin et al., 2012) using the following plugins; MorphoLibJ (Legland et al., 2016) References Legland, David, Ignacio Arganda-Carreras, and Philippe Andrey. "MorphoLibJ: integrated library and plugins for mathematical morphology with ImageJ." Bioinformatics 32, no. 22 (2016): 3532-3534. Schindelin, Johannes, Ignacio Arganda-Carreras, Erwin Frise, Verena Kaynig, Mark Longair, Tobias Pietzsch, Stephan Preibisch et al. "Fiji: an open-source platform for biological-image analysis." Nature methods 9, no. 7 (2012): 676-682.