Metaspectral, a software company advancing computer vision and remote sensing using deep learning and hyperspectral imagery, is introducing its Ore Resource Evaluation System (ORES).
The company was selected to present the technology at the International Minerals Innovation Institute (IMII) DEMOday, which took place on 22 May 2024 in Saskatoon, Saskatchewan, Canada.
IMII is a non-profit organisation jointly funded by industry and government. It is committed to developing and implementing innovative education, training, research, and development partnerships to support a world-class minerals industry.
ORES will automate the sampling process of potash, uranium, and other ores, providing comprehensive, real-time analysis of ore composition and quality. The technology uses artificial intelligence (AI) to analyse data from hyperspectral sensors placed along conveyor belts that move the ores, allowing for continuous, non-contact, non-destructive analysis.
“Our integrated software platform can provide immediate information to operators about ore quality and composition. This can guide early decision-making in the milling process and make it possible to identify and select only ores of a predetermined grade for processing,” said Francis Doumet, CEO and Co-Founder of Metaspectral. “Enabling the selective processing of ores makes it possible to reduce costs and lessen environmental impact, using less water and energy, while producing fewer tailings and less waste.”
This comprehensive level of analysis is not possible using traditional methods of ore sampling, which only analyses a single point on the sample. ORES, conversely, captures complete data about the materials when the ore passes by the spectral sensors on a conveyor belt. These sensors capture hyperspectral data, which measures photon interactions to produce unique spectral signatures that can be interpreted to uncover detailed information about the properties of the ore at the molecular level.
“Fine-grained variations in mineral composition indicating ore grade, impurities, or environmental contaminants can be identified immediately, including the identification of complex, overlapping mineral signatures,” added Migel Tissera, CTO and Co-Founder of Metaspectral. “Hyperspectral images capture enormous quantities of data, gigabytes per second. We use AI not only for the analysis of the data but also to compress the data to just 30% of its original size without compromising integrity. This enables us to stream our analysis at high speeds and provide real-time insights.”
This high level of detail has significant potential to improve the mining sector’s operational efficiency and profitability while lessening its environmental impact. This technology can also help to reduce the need for human exposure to ores which can enhance worker safety.
Metaspectral’s technology has been validated and is already deployed commercially in other sectors, including plastics recycling. A similar conveyor belt configuration has achieved identification accuracies exceeding 92% at high speeds for difficult-to-sort materials such as thin plastic film, black material, and transparent material by polymer type.