Current corrosion assessment techniques are stuck in the past across many industries. Despite many businesses adopting newer image capture technologies, like drones, the bulk of the assessment is still deeply rooted as a manual check. As businesses collect more and more data, they need to evolve their data processing techniques.
Our corrosion detection service (Deepsteel) offers an effective solution that scales to your data rapidly and accurately. It can detect the presence and extent of corrosion with pixel-level granularity. Where an analyst can only go over a handful of assets per day, Deepsteel can process thousands.
Our advanced AI pipeline is composed of 3 stages:
1. Separate the structure from any background to reduce interference
2. Detect corrosion on the structure
3. Localise the corrosion and determine its extent throughout the structure
How good is it?
Deepsteel is soon to be fully integrated into National Grid’s corrosion assessment flow. It detects 78%* of corrosion, which is on par, and in some cases better, than an analyst manually conducting an assessment.
Moreover, it produces a more objective assessment than two or more analysts looking at the same asset.
* 78% on steel towers; accuracy will depend on use case
Will Deepsteel work for me?
Deepsteel learns to understand what your assets look like and more importantly what the corrosion on your assets looks like, by looking at the imagery you have collected.
What kind of data do I need to have?
Our AI model learns best from high-resolution images, in decent lighting and weather conditions. The general rule of thumb is: if a human can identify it, Deepsteel can learn to do so too.
To be able to localise the corrosion, we may also require positional data associated with the images (GPS coordinates) and a LiDAR scan of the asset.
