Traditional draft survey methods rely heavily on human observation from small boats, introducing safety risks and subjectivity into measurements. GOTEC's AI-powered approach fundamentally changes this paradigm.
Our semantic segmentation neural networks can separate waterline and scale digits with remarkable precision, computing accurate draft values even in complex sea states. The system handles wave-resistant curve fitting and perspective tilt correction automatically.
Technical Breakthrough
The algorithm achieves millimeter-level accuracy by combining multiple neural network architectures. One network identifies the waterline boundary, another recognizes scale markings, and a third performs perspective correction. The results are then fused using a calibrated physical model of the vessel.
This approach has been validated across hundreds of vessels in various port conditions, consistently delivering results within acceptable tolerance for commercial settlement purposes.