UK and European materials-sector collaborators

Materials and Industrial Technology

Testing computer vision for material classification

A prototype for distinguishing material categories from images

Period

2025

Scale / geography

Cross-market experimental image dataset

Work type

Proof of concept

Status

Prototype

Market

United Kingdom and Europe

Engagement

Computer-vision prototype

Challenge

Materials-sector collaborators wanted to test whether image-based classification could support faster, more consistent identification under real-world variation.

What we did

We prepared an experimental image set, evaluated model approaches and built a prototype that exposes confidence and failure cases for expert review.

Outcomes

  • The prototype demonstrates candidate classification approaches

  • Error patterns and data limitations are visible to domain experts

  • No production performance claim is made