Back to Projects

Grain Boundary Analysis

Adoptable to your company — custom models, your datasets, your KPIs

Machine-learning-assisted grain boundary analysis for metallurgical samples—not fully automatic by default, but computer-supported throughout. Download a sample PDF report below to review typical deliverables.

Run on local workstations, your own servers, or in the cloud—aligned with your IT policies. Optional automated evaluation where it fits your process; embed the same capability via API.

Download sample report (PDF)

Not limited to this use case

We adapt this solution to your imaging and quality goals

Click or press the arrow to expand examples and the full description.

Grain boundaries here are one reference deployment. MetricVision can train and operate models on your own images and labels—different materials, modalities, and readouts. Tell us what you need to quantify; we align architecture, training data, validation, and reporting with your process.

  • Powder and particle characterisation: size distribution, shape, and agglomerates from optical or SEM micrographs
  • Porosity and void structure: pore fraction, size, and connectivity in cross-sections
  • Surface and coating defects: scratches, pits, delamination, or contamination on metals and films
  • Phase, inclusion, or crack segmentation when the task is not grain-boundary tracing
  • Other microscopy or industrial vision tasks—your specification drives the model and the export format

From manual effort to scalable analysis

The challenge

  • Manual grain boundary analysis is slow and ties up expert time.
  • Subjective interpretation increases error risk and audit friction.
  • High sample volumes strain resources and delay downstream decisions.

The solution

  • Computer-assisted analysis accelerates evaluation; fully automated batch runs are available when your process allows.
  • Reproducible pipelines and metrics improve precision and comparability—with review steps where you need them.
  • Integrate via API into your own software, or operate standalone—from single specimens to higher throughput.

Before and after comparison

Before
After
Interactive comparison: original microstructure versus computer-assisted grain boundary overlay—move the slider to explore.

How the system works

Microscopy images are processed with ML support: structures can be segmented, grain boundaries highlighted, and metrics derived for statistical evaluation. Automation is configurable—from guided, review-friendly steps to automated evaluation where appropriate. The functionality can also be exposed as an API so you embed it in existing tools and workflows.

See the workflow in action in the short walkthrough below.

What you get

Computer-assisted grain boundary analysis; optional automated evaluation

Reproducible metrics and transparent, review-friendly workflows

Detailed statistical evaluation

API integration and embedding in existing workflows

Local, dedicated server, or cloud execution

Data protection and full control over your data

Ready to shorten your analysis cycle?

Share your imaging challenge—we tailor models, training data, metrics, and deliverables to your team.