Pavement Crack Image Detection based on Deep Learning

Abstract

Crack is one of the most common road diseases. Once it appears, the quality of road engineering will be greatly reduced and even cause road collapse. If the cracks can be found in the early stage of timely maintenance, it will greatly save maintenance costs. However, the range of image cracks on the actual pavement is too wide, the image clarity is not enough, the composition is complex, and direct detection is very difficult. The traditional manual detection method takes too long time, has not enough precision, high risk of detection operation, and has a series of shortcomings. Therefore, according to the characteristics of pavement cracks, this paper adopts an automatic detection method based on deep learning. The method first preprocesses the crack image, and then inputs the preprocessed pavement image into the convolution neural network (CNN) model for detection. Experimental results show that this method is accurate and can better detect pavement cracks.