基于卷积神经网络的高楼外墙裂缝检测系统

Convolution Neural Network-Based System for Detecting Cracks on Exterior Wall

基于卷积神经网络的数学模型,通过无人机拍摄外墙图像建立数据库,本文结合软硬件建立了一种外墙及饰面材料的裂缝检测系统,能有效地识别外墙裂缝的严重、一般或轻微三种毁坏程度,且有效识别率分别为86%,91%,97%。

Based on the mathematical model of convolutional neural network, the database is built by taking images of external walls by unmanned aerial vehicle;a crack detection system for external wall and its facing material is established by software and hardware. The system can effectively identify the severity, general or slight damage degree of external wall cracks, and the effective identification rates are 86%, 91% and 97% respectively.