The Lenaeroproekt institute has integrated an AI-based photogrammetry service developed by Cloud X to analyze airport runway surfaces. The system automatically detects and classifies pavement defects, allowing engineers to review the findings. This implementation has reduced the processing time for a single runway from 10 working days to just one.
The AI model, trained on the CX AI Platform, performs pixel-level segmentation to identify cracks and joints. It distinguishes between actual damage and surface features by utilizing a dataset that includes rare defects. The system processes high-resolution images of 10,000 by 50,000 pixels, with each image file starting at 1 GB in size.
Engineers interact with the data through a browser-based interface that supports zooming and manual adjustments. The platform automatically calculates the total length of defects, generates reports, and maintains a historical database for comparative analysis. The final assessment of runway conditions remains under the control of the institute's specialists.