Normalized claim
Accuracy: 84%
Real-world testing achieved about 84% accuracy.
E.ON remote smart-meter diagnostics using Amazon Textract
Use case typeIntelligent document processingUpdated Jun 13, 2026
E. ON Smart Energy Solutions in the UK built a remote diagnostic workflow for smart meters using smartphone video, Amazon Textract, and custom heuristics to detect LED labels and pulse patterns. Customers record a 7-second video in the E. ON app, the system extracts frames, reads printed LED labels with Amazon Textract, localizes LED regions above the labels, counts pulses over 7 seconds, and maps the result to meter error codes with a natural-language explanation.
- Organization
- E.ON
- Industry
- Energy & Utilities
- Location
- Germany
- Published
- June 2025
Reported outcomes
84%
accuracyQuality & accuracy
95%quantified impact
Strategic outcomes
Speed & agilityEnabled remote smart-meter diagnosticsRisk & complianceImproved diagnostic consistencyNew product / capabilityAdded video-based meter troubleshootingCustomer experience & trustReturned natural-language diagnostic explanations
Why do we believe this?Outcome claims, sources, and evidence checks
Normalized claim
Quantified impact: 95%
It helps E.ON maintain its 95% smart meter connectivity target.
Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks
- Customer
- E.ON
- Provider
- AWS
- Maturity
- Unknown
- Linked source
- AWS Machine Learning Blog
No explicit deployment-stage evidence found.
Customer identity supportedSource describes one deploymentMaturity evidence evaluated
Primary read
Use case focus
Showing 3 of 3
- 1Computer Vision
- 2Predictive Maintenance
- 3Document Processing
- Diagnose smart-meter communication/WAN LED errors without costly on-site visits.
- Reduce human error and inconsistency when engineers interpret LED blink patterns.
- Address 135,000 annual appointments and more than £20 million in site-visit costs.
- Customers record a short 7-second smartphone video of the smart meter in the E.ON app.
- The application extracts key frames and uses Amazon Textract to detect and read the printed labels SW, WAN, MESH, HAN, and GAS.
- The detected label positions act as landmarks to localize the corresponding LED regions above each label.
- Custom signal-intensity and brightness heuristics determine whether each LED is on or off in each frame.
- The system counts LED pulses across the 7-second clip and maps the pulse pattern to the meter's error codes.
- The diagnostic result is returned to the customer or engineer with a natural-language explanation.
Technologies
- The solution can diagnose about 350 cases per week, or 18,200 annually, remotely.
- Real-world testing achieved about 84% accuracy.
- The approach reduces unnecessary site visits and improves diagnostic consistency.
- It can detect malfunctioning meters before issues escalate.
- It helps E.ON maintain its 95% smart meter connectivity target.
Architecture
A customer smartphone video is uploaded through the E.ON app to AWS, split into frames, filtered by a signal-intensity heuristic, analyzed with Amazon Textract to read meter labels and locate the associated LEDs, and then processed with custom heuristics to classify LED pulse frequency and map it to meter error codes and a natural-language explanation.
Sources & evidence1
Evidence: Low35/100Evidence strength
- Customer explicitly identified
- Quantified outcome available
- Technical implementation details available
Type: Blog PostPublished: Jun 10, 2025Publisher: AWSEvidence: VendorConfidence: Medium
AI-generated summary. Verify important details with the linked sources before relying on this case.
Explore related AI use cases
Customer evidence
Provider evidence
Market context
AI patterns
Was this useful?
Community
Comments
No published comments yet.