Fitz Coatings.

Assessing corrosion is skilled work. When inspecting a surface, professionals may need to estimate the percentage affected by corrosion and use that to judge its severity. Consistency is the difficulty. Visual estimates vary between individuals, which makes a repeatable assessment hard to achieve. Fitz Coatings saw a chance to explore whether technology could support professionals with a more objective method of analysing corrosion.

The concept was a tool that could analyse a photograph, identify areas of corrosion, and calculate the percentage of the surface affected. That could then be considered alongside RI ratings to give users a suggested guideline for the level of corrosion shown.

The potential went beyond specialist assessment. The same technology could also become a teaching tool that helps people learn to recognise and evaluate different levels of corrosion.

Before investing in a full product, Fitz Coatings needed to know whether the idea was technically achievable.

Key challenges included:

• Identifying corrosion accurately within photographs.
• Accounting for different surfaces, conditions, and lighting.
• Calculating corrosion as a percentage of the overall surface.
• Understanding where AI detection could be relied upon and where human input would still be required.
• Exploring how RI ratings could support the assessment.
• Establishing the technical limitations of the concept before full development.
• Creating a practical roadmap for taking the idea from research into a working product.

We approached the project as a focused R&D exercise instead of moving straight into development.

Working with Fitz Coatings, we explored the proposed user journey and investigated the technical, analytical, and algorithmic approaches required to turn a photograph into a meaningful corrosion assessment.

Our proposed approach combined automation with professional judgement. AI-based image analysis could initially identify and highlight areas likely to contain corrosion. The user could then inspect the image in greater detail, zooming into specific areas and manually refining the selection where greater accuracy was required.

Once the affected areas had been identified, the proposed system could calculate the percentage of corrosion across the supplied surface area, providing a more consistent guideline for assessment.

A proof of concept let us test the viability of this approach without investing early in a full production application. The focus was on function over appearance, which helped us understand what could be achieved, where limitations existed, and which areas would need further development.

The resulting Proof of Concept Document captured our findings, the proposed technical approach, the likelihood of success, known limitations, and the steps required to progress towards a complete corrosion detection tool.

It gave Fitz Coatings a development blueprint grounded in research.

The project gave Fitz Coatings a clearer understanding of how its corrosion detection concept could become a digital product, before committing to the cost and complexity of full development. Through discovery, technical investigation, and proof-of-concept work, we assessed the feasibility of the proposed technology and documented a practical route to implementation.

The R&D phase delivered:

• A feasibility assessment for image-based corrosion detection.
• Investigation into AI-assisted image analysis.
• A functional proof of concept focused on the core assessment process.
• A documented approach for calculating corrosion coverage.
• Exploration of combined automated and manual corrosion identification.
• Identification of technical limitations and factors affecting accuracy.
• An assessment of the concept's likelihood of success.
• A proposed technical and algorithmic approach.
• A development blueprint for a future production-ready application.

Fitz Coatings gained the evidence and technical direction to make an informed decision about future development.