Pavement Distress Detection Based on Transfer Learning

Abstract

With the rapid development of highway construction in China, more and more attention has been paid to highway maintenance. The traditional manual detection and recognition methods cannot meet the needs of highway development, so the research of detection and recognition technology based on road image has become particularly important. In recent years, deep learning has shown very high performance in target detection. Based on transfer learning, this paper reuses part of the network of pavement crack detection based on Faster R-CNN to improve the performance of pavement distress detection.