Capture the material
Images of extracted rock or exposed rock walls provide the material for analysis. Lighting, dust, image quality and the visible surface affect what can be examined.
Technology and R&D
Another view of
the material.
Archon Mining’s technology program explores how computer vision can assist field inspection. Opal AI highlights possible signs of potch or precious opal for experienced miners to review.
Vision
Computer vision experiments supporting opal and material recognition.
Field
Tools assessed against practical site conditions and operating needs.
Review
Potential detections checked by experienced personnel.
Learn
Field imagery and observations inform further development.
Program purpose
From field image to human inspection.
Opal AI is an advisory tool. Its outputs require manual verification by experienced personnel and do not guarantee detection, discovery or commercial success. Field trials help assess where it can usefully support the crew.
Highlight possible signs
The model marks image regions that may contain potch or precious opal. These are potential detections for review, not confirmed finds or measurements of value.
Check in the field
Experienced personnel inspect the material and consider its geological context. Decisions about further investigation remain with the team.
Practical applications
Rock piles, rock walls and changing light.
Project material from 2024 documents Opal AI examples using rock, potch and precious opal imagery, including images captured under UV light. These examples illustrate the intended use; they are not a published accuracy benchmark.
- Review images of material brought to the surface during drilling.
- Examine visible features in images of exposed rock walls.
- Compare observations under different lighting conditions, including UV-assisted inspection.
- Use field images and human review to inform further model development.
Understanding the limits
Human judgement guides the next step.
Opal AI supports visual inspection. It does not establish the extent, grade or commercial viability of a deposit, and it cannot replace direct examination or project assessment.
Detections need checking
Image analysis can miss material or flag something that is not opal. A model indication must be checked against the actual material.
Confidence is not value
A displayed confidence score is a model output. It is not a measure of opal quality, deposit size or the likelihood of a profitable project.
Results are not guaranteed
The examples shown do not demonstrate a guaranteed improvement in detection, recovery or returns. Field evaluation guides further development.