Blade inspection from drone photos.
An offshore wind manufacturer.
98%Accuracy on blade photos
- 90%Less photo-checking time for engineers
- 70 to 80%Accuracy of the earlier AI system
- On siteRuns without a cloud connection
- Customer
- An offshore wind manufacturer
- Industry
- Energy, construction & manufacturing
- Deployment
- On site, offline
- Workflow
- Visual inspection
BEFORE MATRIX
The challenge.
Inspecting offshore turbine blades means sending crews out to sea. An earlier AI system was only 70 to 80% accurate on drone photos, so many trips were spent confirming damage.
THE WORKFLOWWhat we built,
What we built,
step by step.
A workflow in steps, each with its own model: flag possible damage, outline it, check its depth and grade its severity with a confidence score. Low-confidence photos go to a blade engineer, who decides whether a crew goes out.
- 01One model flags possible damage on each drone photo.
- 02A second outlines it, and a third checks its depth against dirt or shadow.
- 03Each finding is graded for severity, with a confidence score.
- 04Low-confidence photos go to a blade engineer, who decides whether a crew goes out.
UNDER THE HOOD
How it works.
Erosion
Let’s make AI useful.
For your actual work.
Tell us about a job your team repeats, not your confidential files. We’ll show you how Matrix handles work like it.
Request a demo