Deep Learning Comes to Thin Sections
The full capabilities of today’s machine learning are now available for thin section image interpretation. The SubsurfaceAI custom deep learning application for thin sections analyzes entire images in a highly efficient workflow through an intuitive user interface.
Geoscientists can easily label, annotate, and interpret a small subset of the thin section data. Customized AI/machine learning models learn from the geoscientist, providing suggestions and analyzing the entire image. Trained models can be used to quickly analyze all similar thin sections.
Results of the full segmentation of the image are presented in an interactive visualization, where each individual grain has been identified, along with the pore space, clay, and any other material present. The AI-based results are analyzed to extract shape and size distributions, presenting these in customizable plots that can be filtered and used to present statistics.
Virtual Core integrated geoscience visualization and analysis application
The Virtual Core application capabilities in AI/machine learning enable extracting features for classification, eliminating the drudgery of core analysis while providing a deeper understanding.
Virtual Core Features
Data Visualization Capabilities
This deep dive demonstrates the Virtual Core application capabilities in visualizing integrated data sets across multiple scales to create a system that is intuitive. The system enables the user to develop stronger insights, in particular for subsequent AI and machine learning.
AI And Machine Learning Techniques
This deep dive demonstrates the Virtual Core application capabilities in AI and machine learning techniques for extracting features and classification. The Virtual Core application eliminates the drudgery of core analysis while providing a deeper understanding of the data.
CT Scan and Photograph Data Preparation Service
CT Scans and core photograph images are underutilized due to the difficulties in cleaning, preparing and integrating with other data. Virtual Core corrects CT data and integrates with core photographs so that CT data can be re-evaluated and used more frequently.
Featured Case Studies
Applying Machine Learning to Subsurface Data Reduces Uncertainty
Analyzing and upscaling cores and other subsurface data using AI/machine learning techniques creates a new level of efficiency and understanding for geoscientists.