SwitzerlandLogistics and transport2017 – ongoingSBB (Swiss Federal Railways)
Deep learning for rail-defect inspection
SBB and CSEM reported a 10-fold improvement in rail-defect detection in field tests
Problem
Manual track inspections became harder as train traffic increased. Earlier image-analysis methods generated too many false findings because outdoor rail conditions varied widely.
Approach
SBB worked with CSEM on software for images collected by diagnostic trains. CSEM developed modules to detect and classify anomalies, with rail experts supplying labeled examples. Human specialists still checked the automated findings.
Results
No published figures
Takeaway
The software supported inspectors handling large image collections. Expert review remained necessary because automated analysis had not matched human assessment quality.
Categories
- Size
- Large enterprise
- Industry
- Logistics and transport
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Computer vision
- Adoption
- Partner co-development
- Evidence
- Company's own claim
Sources
Source summaries
SBB and CSEM staff described deep-learning software for diagnostic-train images. They explained the false findings produced by earlier methods and the continued need for manual verification.
Full details are in the original sources.
