Research on Concrete Disease Image Recognition Technology Based on Pre-Training Neural Network
During the service life of concrete structures,various factors can lead to different degrees of damage or defects,including cracks,leaks,carbonization,steel corrosion,and concrete spalling,which seriously affect the service life and safety of concrete structures.Due to the low reliability,high cost,and strong subjectivity of manual crack detection methods,this study explores the use of pre-trained neural networks for image recognition of concrete defects.This method utilizes existing pre-trained networks,which only require importing healthy concrete images and activating corresponding neural network feature layers to calculate the statistical thresholds of large samples in healthy conditions.By comparing the statistical thresholds of the test set,it can classify and label the abnormal regions of defect images.Compared with traditional methods,this method has the advantages of high accuracy,low influence of human experience,high flexibility,fast speed,and low cost for image recognition of concrete defects.
NETL
NSTL
万方数据
