ARTGoldTM, ARTPhot, ARTSection, ARTMorph, ARTMin-I, ARTMin-II, ARTMin-III, ... We mastered the art of automated mineral recognition... But it's not enough, and look forward...

In collaboration with Dr Paul Bédard of the Applied science department and Dr Kevin Bouchard and Dr Julien Maître of the Mathematics and computer science department of Université du Québec à Chicoutimi (UQAC), and subsidized to the level of 300,000$ by the Québec's Nature and Technology Research Fund (FRQNT) and Québec's Energy and Natural Resources Department (MERNQ), IOS initiated a 3 years research partnership on "Machine learning based automated optical reconnaissance of mineral".

That's geologist jobs: Identify minerals, identify rocks, identify structures and textures. And if human eyes and brain can recognize these, so shall the machine! The proof of concept was done and published (https://doi.org/10.1016/j.cageo.2019.05.009) a year ago, and we managed to process images in order to properly identify minerals from RGB pictures. We now aim to make it operational, robust and usable. It's a huge challenge, but applications could be as disruptive: automated recognition of detrital mineral, detection of indicator minerals, quantification of industrial mineral, pebble counting, rock identification in hand sample or outcrops, and ultimately... core logging without the use of complex instruments. And these are only one of the options we are exploring... The dream: Relieving the geologist of routine data acquisition tasks, and using their skills on interpreting the data!

It's R&D, so it's uncertain... we are dreaming, but we are capitalizing on our track record! The team is smart!

‘'There is those who follow the rules... and those  who define them''

RGB image of silty material taken with a binocular microscope, with various other images issued from numerical processing, enabling automated mineral identification.