Development of a Framework for Identifying Concrete Pavement Cracking Distresses Using Machine Learning
Building on previous research using drone-based pavement data collection and public-submitted pothole photographs, this project will develop an automated approach for evaluating pavement conditions using artificial intelligence (AI), machine learning, and computer vision. The research team will train advanced image-analysis models using an extensive database of pavement images and videos representing different roadway conditions, pavement distress types, and environmental settings. The resulting system will be designed to automatically detect, classify, and quantify common pavement distresses, including cracking, rutting, potholes, raveling, and patching. Where feasible, the system will also assess distress severity and calculate measures such as crack length and density, pothole dimensions, affected pavement area, and rut depth. By replacing time-intensive manual image interpretation with automated analysis, the research will provide a scalable approach for network-level pavement condition assessment.
California State University, Chico
The automated pavement assessment methods developed through this research could be incorporated into transportation agency pavement monitoring and management practices. Agencies could use imagery collected by drones, roadway surveys, or public reporting systems to rapidly identify and quantify pavement deterioration. The resulting data could also be integrated with pavement management systems and performance models to help agencies prioritize maintenance and preservation activities.
Implementation could substantially reduce the time and staff resources required to assess pavement conditions while improving the consistency and objectivity of pavement evaluations. Faster processing of large volumes of roadway imagery could allow agencies to monitor pavement conditions more frequently and across larger networks. More timely and consistent condition data could, in turn, support better maintenance decisions, more effective allocation of limited resources, reduced pavement management costs, and improved long-term roadway performance.
-
Contact Us
San José State University One Washington Square, San Jose, CA 95192 Phone: 408-924-7560 Email: mineta-institute@sjsu.edu
