A Large-Scale Image Repository for Automated Pavement Distress Analysis and Degradation Trend Prediction
Hanlin Yang, Jinpu Cao, Jun Wan, Qian Gao, Chenglong Liu, Martin Fischer, Yuchuan Du, Yishun Li, Pooja Jain, Difei Wu
Abstract
In recent years, automated detection technologies for large-scale pavement distress have become a focal point of research in the transportation sector. With the rapid advancement of deep learning technologies, data-driven artificial intelligence recognition algorithms have gradually emerged as the industry mainstream. The effectiveness of such algorithms largely depends on the reliability and quantity of the samples. However, existing datasets exhibit significant shortcomings in terms of sample size, category diversity, and support for distress tracking. In this study, a large-scale image dataset was meticulously constructed. This dataset includes 51012 road images for pavement distress identification and 8928 images for long-term tracking of pavement distress. Using this dataset, six mature object detection algorithms were trained and evaluated, with the results demonstrating the performance of these algorithms. To the best of our knowledge, this is the first large-scale pavement distress dataset that includes long-term tracking of pavement distress, providing reliable data support for dynamic tracking and monitoring of pavement distress as well as for optimizing road maintenance strategies.