Shawn Hood

Chief Technology Officer, GoldSpot Discoveries Corp.

Shawn is a P.Geo in Economic Geology with experience across a broad base of previous exploration and mine geologist roles ranging from open pit, underground, brownfields and greenfields projects in Canada and globally. Shawn applies machine learning techniques to understand ore deposits by integrating disparate datasets and enhancing mineral exploration work programmes. He has a focus on understanding how this technology can deliver objective and repeatable results in mineral exploration settings, especially where trained geoscientists can apply their expert knowledge to guide machine learning models, and situations where data-driven results can be field-validated and improved by geologists. Shawn is a graduate of Carleton University in Ottawa (BSc Hons.) and the University of British Columbia’s Mineral Deposits Research Unit (MSc). His PhD research demonstrated the application of machine learning towards mineral exploration and investigated why geologists’ input is fundamentally important.

Abstract

Automated Logging using Machine Vision: Using Large Datasets to Improve Geological Models

Machine learning and computer vision techniques are increasingly being used to standardize descriptive processes in geology, such as drill core logging. The result is a rapid increase in data quantity and quality. The presented case study demonstrates how a large number of drill core images can be processed using artificial intelligence and cloud computing methods to extract geological data. The desurveyed results are represented in three dimensions (3D) and used to produce a block model for download by end users, e.g., geoscientists, technicians or engineers. The discussion of automated logging benefits in the mining industry tends to focus on productivity: the efficient use of personnel and the related time/money savings; however, the rapid creation of consistent geoscientific information can reduce risk by creating better decision-making tools, namely, geological models. Improvements can be observed at the drillhole scale (precision and accuracy of automatically extracted information versus manually logged information) and in a holistic macro scale (comparing models made from large datasets to earlier models made using insufficient data to represent geological features).

Speaking at

Theme Session: Engage. Connect. Evolve.

February 1, 2022 @ 1:30 pm - 3:30 pm PDT