Russia Develops AI System to Detect Cracks in Buildings
Researchers across the country are building computer vision algorithms capable of spotting structural defects early, aiming to prevent accidents and improve construction safety.
Experts at North-Caucasus Federal University (NCFU) are developing a system that can identify cracks in images of building structures. The technology is based on computer vision algorithms and methods of continuous model retraining.
Currently, the model is being trained to classify crack types and identify their causes using photographs. Future upgrades will expand its capabilities to include video diagnostics.
The research team in Stavropol is targeting a diagnostic accuracy of no less than 85%. The final acceptable error margin will be determined after field testing.
A crack recognition system for concrete structures was previously developed at Novosibirsk State Technical University (NSTU), achieving 95% accuracy. Meanwhile, Perm National Research Polytechnic University (PNRPU) is working on a similar AI-powered program to detect hazardous buildings.
