DimEye - AI Laser Line Detection for Subsea 3D Inspection
Measuring a mooring chain underwater is harder than it sounds. Pull it to the surface, or send divers and ROVs down to measure it by eye. Either way it is slow, and the result depends on who is looking.
DimEye Corp provides Video Laser Scanning (VLS) services that build high-resolution 3D point clouds of subsea mooring chains and other critical assets while they stay in place. Inspection teams get measurable data instead of estimates.
Those models are only as accurate as the laser line position in every video frame. DimEye wanted an application that could find each laser line automatically, classify it as straight, curved, or over-exposed, and apply a localization method tuned to each type, for faster and more accurate results.
Working with DimEye, Numurus built that application on NEPI, running on a Numurus S2X with an NVIDIA Jetson Orin AGX. NEPI provided the image processing interfaces and the remote browser-based interface, so both engineering teams could focus on the parts unique to DimEye: training the AI classifier and the per-type localization methods.
A working prototype was delivered within 3 months and refined with DimEye's feedback the following month. Numurus engineers also helped DimEye's IT team set up a remote portal so the application can be reached from anywhere.
The work is ongoing. Numurus continues to optimize the processing for each laser line type as DimEye runs more data through the system.