Sales :  Support :

Registered in England and Wales company # 03449181.
This website does not use cookies.

© Copyright: Conwy Valley Systems Limited  2000 - 2026

Conwy Valley Systems Limited,

West Acre, Llys Helyg, Deganwy

CONWY LL31 9BN, United Kingdom.

VAT# GB 741 5190 48

A Geoscience Wales  founding member.

PETROG includes patent protected technology.

With the advent of both fast hardware and mature software, AI has developed into a toolkit that can help in complex tasks.

We believe we are a long way from being able to interpret petrographic images automatically, at least to the level of detail that highly trained geologists / petrologists / petrographers can collect using PETROG.  However, AI techniques can provide assistance to these experts to take away the mundane parts of a job, and allow these experts to maximise their skill set.

One aspect that PETROG can assist with is to allow you to collect data and then perform analysis of that data in a structured form.  State-of-the-art Machine Learning approaches to mineral identification increasingly rely on multimodal, spatially referenced petrographic datasets, where each labelled point is tied to its exact location and corresponding imagery. PETROG’s structured data model - linking compositional labels, accurate point locations, and saved microscope images - aligns directly with the requirements of recent AI research.

The 2025 LITHOS dataset paper notes that automated petrography requires “the largest and most diverse publicly available experimental framework” to achieve robust mineral classification, and stresses that expert annotation is essential for training reliable models [LITHOS 2025] . Similarly, the 2025 Scientific Data photomicrography dataset emphasises that “precise location information for each labelled region” is critical for learning mineral boundaries and textures [Scientific Data 2025]. Broader mineral-ML research echoes this: the 2025 MRMinerals dataset paper highlights that AI in mineralogy depends on “machine-readable, consistently structured training data” to support reproducible model development [MRMinerals 2025].

PETROG’s ability to generate large, spatially explicit, expert-verified datasets therefore provides an ideal substrate for supervised learning, segmentation models, and multimodal architectures aimed at improving automatic mineral identification.

AI and Machine Learning

AI/ML Publications and Conference Highlights