ALROSA has connected artificial intelligence to the sorting of diamond-bearing rock at its alluvial deposits. A machine vision system, developed together with Yakutia’s AI Laboratory, tracks the movement of stones and singles out the most valuable ones. Several exceptionally large specimens have already been found at the Mayat placer mine.
A camera continuously films the flow of rock on the conveyor belt. The neural network breaks each frame into separate objects and evaluates each one, filtering out waste rock and fakes. According to Tatiana Platonova, head of the AI Laboratory, the model is trained on extra-large diamonds, and in testing it recognized more than 96% of imitations.
The first site was the Mayat placer mine in the Anabar ulus, where the project was launched last year. The technology is being used at deposits operated by ALROSA’s subsidiary Almazy Anabara.
Automation is not limited to mining. At the Yakutsk diamond trading enterprise, twenty machines built on solutions from the subsidiary Burevestnik Engineering Center have been in operation since 2018. The machines independently determine the color, carat weight, shape, clarity, and quality of stones.
Albert Kapitonov, leading expert at the sorting shop, explained how the equipment is trained: before launch, the machine is run through thousands of digitized photographs and video recordings. Only after accumulating a sufficient sample base does the automated system begin to distinguish categories with confidence.
ALROSA is consistently removing manual labor from the diamond recognition process. Neural networks at the placers and automated lines in Yakutsk reduce the risk of losing a valuable stone and speed up processing. For the industry, this is a clear example of machine vision displacing traditional inspection.
Source: ALROSA
