/ technology

Computer Vision, the Chocolate A-Ice Cream

What is Computer Vision?
Computer Vision is a type of Artificial Intelligence. If AI is ice cream, Computer Vision would be chocolate ice cream. Strawberry fan? Try audio data like voice recognition...

For a fun read on how creating Artificial Intelligence is similar to creating a recipe for your favorite food, check out our article Bite Size AI!

Computer Vision is the process of "teaching" the AI to identify visual data. Each image contains millions of pixels of data that an AI can be programmed to interpret. Data like color and intensity, which can be related to physical data like depth, absorption, or reflection. The proliferation of sensors to detect and record the various forms of data have led to an explosion of data available for mining. Most of these sensors are able to collect data beyond the normal human experience- for example infra-red light or being able to slow down or stop time- capturing and retaining the data of an explosion one frame at a time.

Visual Data (images) can be gathered by a variety of techniques like

Light-Spectrum-image--800-wide Image from ArcGis

The kinds of images that the AI can derive information from are legion. If you can see it, the AI can be taught to see it as well.

Churning the A-Ice Cream
AI, like Ice Cream, doesn't magically appear, it has to be created through a laborious process of adding together different ingredients, cooling to the proper temperature and churning and churning and churning until its just right...

AI is a set of mathematical algorithms created by humans. These algorithms are then applied to a "training data set" to see if they yield the expected results. Essentially, the algorithms are programmed to find certain rock crack characteristics in a data set. They are shown images of rock cracks repeatedly and tested to see if they can find the characteristic patterns in the image data. If they cannot or do so with a poor success rate, the algorithms are refined and tested again.

The AI model is constantly being tested and refined with new images. Once the algorithm is able to accurately pick out a crack instead of a shadow (perhaps the shadows edges are 1 dimensional) or a blade of grass (cracks are not green, the grass is in front of the rock not inside it), its fed new images to test further. Big cracks, little cracks, deep cracks, shallow cracks, cracks in light rock, cracks in dark rocks, etc...

As the AI model is tested on more and more images and its mistakes are continually corrected and the AI is refined, it "learns" what a crack is and how to "see" the crack. Essentially, every time the data matches a set of characteristics, it means a crack is present. An AI model can comb through vast amounts of data in milliseconds allowing it to find patterns in data that could escape a human's attention.

Creating the A-Ice Cream Sunday
Much like creating an ice cream sunday, once you have the ice cream scoop it’s time to decide on your toppings. Chocolate ice cream is nice, but when you combine it with fudge, nuts and whipped cream- it's amazing.

Computer Vision AI can be applied to a vast variety of visual data and when combined with other data is able to create incredibly powerful insights like...
Comparing the data in one image to the data in another creating 3D models.

  • Distinguishing spectral bands and matching them to their mineral using hyperspectral imaging data
  • Distinguishing specific temperatures in thermal imaging and comparing them to other data to determine different gasses burning in a blast
  • Tracking the movement of a rock through the air to predict where it will land
  • Identifying a specific object- like an endangered lizard or a person
  • Identifying visual "noise" like foliage and removing it to "see" the ground beneath it
  • Comparing images of a crack over time to see if its expanding before a slope fails
  • Identifying bedding planes or fissures for blast loading optimization

Time to Dig In!
Because data is available at wavelengths the human eye cannot detect, in quantities too small, or patterns too complex for humans to see, Computer Vision can be an extremely powerful tool in any mining professional's toolbox.
The only thing needed is the vision to see where it can go!


Hungry for More?
There are many AI applications for drilling, blasting, and mining and we love writing about them.

Check out our 2 Free E-books:
AI Guide for Drilling and Blasting
AI Guide for Mining

Watch our YouTube videos