In order to improve the quantitative detection and identification method of building exterior insulation layer defects,based on the defect detection principle of infrared thermal imaging technology,the study analyzes the infrared and thermal image characteristics of cracks and hollow defects with prefabricated defect test,determines the suitable detection environment of two kinds of defects,and proposes a network model that introduces SE-Block and replaces the activation function.The test verifies that the model has 97.89％and 97.25％recognition accuracy for cracks and hollowing defects,respectively,which proves its effectiveness in identifying defects in the external thermal insulation layer of building exterior walls.This study can not only provide technical reference for the maintenance of building exterior insulation layer,but also provide theoretical support for the application of deep learning in defect recognition.
维普
万方数据