Development of 3D Geometric Analysis Neural Networks for Detecting Structural Weaknesses of Buildings Based on Their Visual Representations

This thesis presents a novel deep learning approach for autonomously detecting structural vulnerabilities in urban infrastructure using two-dimensional visual data. By integrating spatial reconstruction algorithms with defect-recognition neural networks, the proposed method transitions flat imagery into semantically analyzed three-dimensional models. The framework effectively identifies topological anomalies, surface deformations, and micro-fissures. The conceptual findings demonstrate enhanced diagnostic precision and computational efficiency, offering a scalable tool for preemptive architectural maintenance and smart city hazard mitigation.

Перевод пока недоступен