Segmentation of Petrographic Images Using Quantitative and Automated Tools

The Department of Natural Resources Canada (NRCan) is seeking to develop automated petrographic image segmentation tools through a sole-source contract awarded to the Institut national de la recherche scientifique (INRS), based on its unique qualification to meet essential technical criteria. The project requires the creation of four key deliverables: guidelines for standardizing and preprocessing thin-section images using open-source tools like Segment Anything Model (SAM); an automated mineral classification system using clustering or deep-learning techniques; integration of these tools into the Analytical Petrography Laboratory workflow with development of a deep-learning system to recognize geological features; and a final report demonstrating methodology, tool performance, and application on complex samples. The contract has an estimated value of $130,000.00, excluding GST/HST, and runs from the award date until August 13, 2027. Only one supplier has been determined capable of fulfilling the mandatory requirements, which include a minimum of 15 years of academic research experience in AI-driven data assimilation for Canadian geological environments, peer-recognized expertise in this domain, and demonstrable experience since 2021 in developing machine-learning algorithms for geological image processing. Intellectual property rights for all deliverables remain with the contractor, and the work is to be performed primarily at the Analytical Petrography Laboratory in Canada using samples from the Geological Survey of Canada. Submissions to challenge the sole-source determination are accepted until August 7, 2026, at 2:00 p.m. EST, and must be sent directly to the contracting authority, Anik Samson, via email or mail. No competitive pricing, line-item structures, or U.S.-style FAR clauses apply; instead, the procurement follows Canadian government procedures under the Government Contracts Regulations and is exempt from full competition under multiple free trade agreements due to the uniqueness of the supplier’s qualifications. No packaging, marking, invoicing details, or COR/COTR information are specified beyond the primary point of contact and estimated contract value.