SulfideNet: Deep Learning for Detection and Quantification of Iron Sulfides in Drill Core Scans
Bibliographic record
Abstract
Iron sulfide minerals such as pyrite, chalcopyrite, and pyrrhotite are among the most critical minerals in mining exploration, yet their precise detection and quantification remain highly challenging at a commercial scale due to the subjective nature of core logging and the inconsistencies that arise from discrepancies between individual geologists' judgement. This paper introduces SulfideNet, a novel deep learning framework designed for fast and accurate segmentation and measurement of sulfide minerals in drill core imagery. Leveraging deep learning, SulfideNet is trained on a curated dataset consisting of thousands of high-resolution core images, each paired with detailed, high-quality binary masks accurately identifying sulfide minerals. SulfideNet was evaluated with multiple validation strategies at various scales. First, it was benchmarked against expert geologist annotations on two validation datasets from orogenic gold and porphyry/epithermal deposits, achieving a high correlation in estimating sulfide mineral percentages with an MAE of 0.41%, and near human-level accuracy in pixel-to-pixel comparisons with Dice coefficient of 82%. Additionally, in prospective studies where geologists evaluated the quality of SulfideNet outputs on 5,664 intervals, it achieved an acceptance rate of 91.9%, demonstrating its reliability and potential for automated sulfide mineral quantification. The results indicate that SulfideNet delivers robust and reliable detection of sulfide minerals, positioning it as a useful AI-assistance tool for geologists in core logging. This innovation leads to improved consistency of core logging, improved geological models, and ultimately more informed decisions in mining exploration and processing.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".