Core photos new focus for AI exploration
GoldSpot unveils LithoLens as newest machine learning tool Metal Tech News – October 21, 2020
Last updated 10/27/2020 at 5:24pm
GoldSpot Discoveries Corp., already known for using machine learning to assist mineral exploration companies, has added new core imaging technology that explorers can use in determining viable drill targets.
This new tech, LithoLens, is a relogging workflow system powered by machine learning to turn drill core photographs into digital images and uses deep learning algorithms to enhance images and extract information.
GoldSpot has recently partnered with Calibre Mining Corp. to create prospectivity maps and 3D targets at multiple projects in Nicaragua.
"The examination of drill core photos is the most anticipated aspect of this engagement, utilizing our proprietary new tool-LithoLens," said GoldSpot Discoveries Executive Chairman and President Denis Laviolette. "We are eager to put LithoLens to work on over 300,000 meters of core to produce fresh and accurate geological logs for El Limon and Libertad to help identify high value drill targets."
An important part of searching for potential lodes of gold and other minerals is drilling a hole into the earth and analyzing the drill core that is brought to the surface.
These core samples are then logged by a geotechnician which typically involves painstakingly laying the entire column of earth on long tables, in a well-lit facility.
The core is then photographed and logged, which is a complete examination of the core.
This process requires the technician to make notes on all the geological sequences encountered in the hole, this can range from rock type, veins, mineralization and many others.
The data from core samples taken from dozens, and often hundreds, of holes drilled into a single deposit begins to provide a 3D rendering of the underground geology and mineralization.
Yet, issues can arise from logging so much core – inconsistencies can arise due to different skill levels of technicians, varying methodologies or incorrect analysis, to name a few.
GoldSpot's LithoLens is designed to provide a non-biased and consistent logging system based completely on the data provided in a photographed core.
"Our LithoLens platform allows us to take old core photos and transform them into intact, geo-referenced digital core images, and then apply deep learning algorithms that enhance these images and extract valuable geological information," Laviolette said.
By uploading photographs from present or historical core photography, LithoLens can extract key information to allow smarter exploration.
The core relogging workflow utilizes powerful and secure cloud-based data processing and includes two distinct deep learning steps.
The first step is automated photo homogenization, where image data is cleaned, processed, extracted, and georeferenced.
The second step is deployment of a deep learning algorithm that can recognize varying geological intervals-or specific features such as veins-within the core.
By finding new use and value from otherwise unused core photography, LithoLens can provide new data for 3D modelling and exploration purposes.
And with the capabilities of machine learning, data models can be trained by knowledgeable and experienced site geologists, allowing core relogging exercises to proceed at speeds which are impossible using the conventional and manual approach.
"At GoldSpot, we believe old core photos are an incredibly valuable dataset, although they aren't being used to add value by most companies," said the company president.
Presently, several mineral exploration companies are utilizing GoldSpot's LithoLens technology at their projects:
• Cassiar Gold Corp. relogged 10,200 meters for quartz vein detection at its Cassiar Gold project in northern British Columbia.
• TriStar Gold Inc. relogged 14,000 meters for cobble detection at its Castelo de Sonhos gold project in Brazil.
• Avalon Investment Holdings relogged 27,300 meters to identify quartz veins at the Omai Gold Mine in Guyana.
• McEwen Mining identified three generations of veins and alteration through the relogging of 136,000 meters of core from the Grey Fox deposit area in the historical Timmins Gold Camp in Ontario, Canada.
Given its ability to help solve geological problems, especially when geological logging is inconsistent, GoldSpot anticipates the delivery of LithoLens as a means to reexplore past projects.
With an absolutely genius move to utilize an otherwise dried up vein, GoldSpot has seemingly created a potential new demand by reexamining old core samples from long thought exhausted avenues and is reinventing the industries approach towards exploration.
With another notch on the belt for the capability of machine learning, GoldSpot charges forward in the diminishing wall between technology and mineral exploration.