Datarock Core is our first product offering centred around extracting information from drill core imagery.

The problem with core photos

Core photography is one of the cheapest and most routinely collected datasets within exploration and mining.

This imagery is a dense source of data, but also significantly under-utilised within the industry. 

Using deep learning we can process the images and extract several important types of geological and geotechnical information.

Our solution

Datarock Core provides a production system for the application of deep learning to core photography and other core imagery types.

Users can manually upload their imagery and metadata through our web portal or set up automatic links to transfer images as they arrive. The platform will perform some QAQC then process this imagery into a cropped and depth registered format ideal for use in various core visualisation software.

Once the imagery is in this processed form, it is ready for further analysis to extract geological and geotechnical features of interest.

Deep learning models - such as those that extract veins or fractures - require training, so we will work with your team to build appropriate training sets. The trained model will be built by us, validated then pushed into production on the DataRock platform. Users will be able to interact and visualise the model results and export output data to the desired location.

shutterstock_396858616.jpg

Core photography models

  • Rotation and distortion corrections

  • Cropping

  • Depth registration

  • Textural domaining

  • Vein segmentation and orientation measurement

  • Fracture detection and orientation measurement

  • RQD estimation

  • Custom segmentation models