AI for asset inspection
and network mapping

1.8M customers across Victoria

Almost 1 million poles in network

Adapted their model for a new use case

Play Video




Powercor/Citi Power

AI Technique

Computer Vision

Electricity distributors are required to regularly inspect their infrastructure and assets. This keeps people safe while keeping the lights on. Powercor saw an opportunity to supplement the manual inspection process with AI-powered inspections using photos captured by aircraft. While this project was in development, the immediate need arose to audit a particular component.

Using HighLighter, Silverpond and Powercor are collaborating to develop a machine learning model to identify assets.

Upon receiving the request to audit the network for a specific make and model of insulator, the machine learning model quickly identified this item in the training photoset. EXIF data from the photos was then used to map its location across the network. This allowed Powercor to find the insulators without sending crews to inspect every asset across the state. 

-Machine learning detects assets with high accuracy
-Seamless feedback mechanism for asset inspection process
-Machine learning unlocked  insights and value from data that may have otherwise been inaccessible 
-Rapidly adapted model to identify and locate a specific make and model of insulator across the network.

Screen Shot 2019-11-07 at 1.44.20 pm
Power pole with selected components identified
Screen Shot 2019-11-14 at 1.04.36 pm
Small objects annotated and detected from areial images 
Screen Shot 2019-11-07 at 1.12.02 pm
objects of interest mapped across the network 

Like what you're reading?
Check out our other case studies below