Leveraging Machine Learning and Open Source Tools for Rock Image Characterization at the Pore-Scale

Executive Summary A software for pore-scale rock image analysis and interpretation has been developed based on open source tools and libraries. The software focuses on specialized workflows and processes for rock characterization rather than image processing workflows and tools. Example workflows implemented include grain morphology estimation, rock component identification, carbonate rock classification, and palynomorph darkness index calculation. Overall, the developed workflows provide quantitative results, improves the accuracy and lowers the bias in the interpretation.