SubSLAM®
Artificial Intelligence
Artificial Intelligence ensures you focus on the asset
Using the latest in computer vision and AI technologies, SubSLAM’s algorithms determine the important features in the scene and ignore the ones not associated with the subject. Focus on the inspection at hand rather than unwanted objects that hinder your field of view.
Boost efficiency and minimise rework
Vision-based reconstructions are hampered if there are fixed features in the scene such as Cathodic Protection probes and manipulators. SubSLAM’s built-in machine learning recognises parts of the ROV and removes them from the reconstruction, ensuring that models continue to be built accurately, without distortion. Multiple tasks can be run while SubSLAM collects data in the background, speeding up acquisition time.
Maximum visual clarity
Underwater assets often become artificial reefs in their own right; good for marine life, but typically a hindrance for visual inspections. Fish and mobile life detectors remove noise from point cloud data for more complete, accurate and less noisy reconstructions.
More meaningful data
Our ML models are trained to classify specific pixels and attribute them to appropriate categories. This gives you semantic information in real time, telling you what you’re looking at and where it’s located. This knowledge feeds directly into Atlas AI while a campaign is underway to make searching for the exact piece of data you need effortless.
Your ML models, our hardware
You know your business better than anyone else. If you already have ML models you can deploy them across the Vaarst suite to get the data insights that matter most to you. SubSLAM is flexible and compatible with different ML models so you can act fast on accurate, meaningful data at the edge or in the cloud.
Get in touch to get started and see how we can help
Contact us