Use case: Asset Inspection

Problem

National inspection of conductors difficult and expensive using helicopters and manual inspection. Increased frequency of inspection and better quality outputs would enable more effective asset management

Solution

Custom machine vision model
- Find power line in image
- Detect damaged section of power line
Cloud processing of images in batch
Present predictions to domain experts to confirm
Machine learning model improves to enable automation

Result

Human: 5 days per 2,000 images
Model Version 1: 10mins per image
Model Version 2: 30-70 seconds per image
Accuracy: >93% Wire defects detected

See other use case

Transport and Logistics

Read use case