Will geometallurgical characterization using machine learning techniques improve models?

October 7, 2019
Posted in News
October 7, 2019 Sam Law

At Imago we believe the answer is a resounding YES.  And it’s more of a “when” than a “will it” kind of thing.  Machine learning is here to stay, and only going to continue to improve in the coming months and years.

An Imago machine learning technique…

The Imago approach to ML has 3 easy steps:

1) Train the model by outlining the mineral of interest in your image

2) Generate the model by hooking this information into a machine learning provider

Sulfide Segmentation Example

Watch the 40-second video below to see this in action

Benefits of this approach

With Imago software, you can do the following easily:

  • You can run it on site and on demand
  • There is no sample preparation needed
  • The cost of running a sample is pennies compared to 1000’s for single QUEMSCAN result
  • The fancier we get in the optics and scanning the higher the cost and the resolution that is possible

The bottom line

Because we can run many more samples with the technique described above, accuracy/precision per sample will be lower, but your accuracy/precision for the deposit will be greater.

Ready To Get Started?


Helpful Links & Information


Guides to get you up & running Imago as well advanced topics like developer info and more.



Coreshed, Connect + Tools, and Imago Mobile. All downloads available from the link below.



With offices around the world, we’re available to help you with any of your needs.


© Seequent Limited, a Bentley Company. All rights reserved