"Leveraging GeoAI for Smarter Transportation Asset Management: From Aerial Imagery to Actionable Roadway Infrastructure Insights"

This event is sponsored by FAMU-FSU College of Engineering Department of Civil & Environmental Engineering.

Abstract: This study presents a GeoAI-based framework for automated detection and mapping of sixteen roadway pavement-marking and traffic-control features using high-resolution aerial imagery and deep learning. Traditional roadway inventory methods are often costly, time-consuming, and difficult to scale, limiting the availability of timely and consistent infrastructure data for safety and asset management. The proposed approach employs a multiclass YOLO-based object detection model integrated with GIS workflows to identify features such as turning lane arrows, crosswalks, bicycle markings, school zones, stop markings, and transit-related symbols across urban and rural roadway networks. A case study from Florida demonstrates how AI-derived detections are converted into georeferenced inventories and validated against ground truth data. Results show strong detection performance across most feature classes, supporting the feasibility of large-scale, automated roadway inventory development. The study concludes by discussing implementation considerations, data requirements, and opportunities for integrating GeoAI-based feature extraction into transportation agency operations.

Richard Antwi, Ph.D.

Geospatial Solution Developer

HNTB Corporation

Speaker Bio: Dr. Richard Antwi is a Transportation Engineer and Geospatial Scientist specializing in GeoAI, intelligent transportation systems, and infrastructure resilience. He holds a Ph.D. in Civil Engineering from Florida State University, an M.Sc. from University College London, and a B.Sc. from KNUST, Ghana. His research focuses on applying artificial intelligence, computer vision, and GIS to roadway asset management, safety analysis, and disaster response. Dr. Antwi has led federally, and state-funded projects supported by FDOT, USDOT, and NSF, and has published extensively in leading journals in GIS, artificial intelligence, transportation safety, accessibility, and infrastructure resilience areas. He has also presented his research at leading professional conferences, including ASCE, TRB, ITE, and IHEEP. Dr. Antwi currently serves as a Geospatial Developer at HNTB, where he develops machine learning solutions, builds data pipelines, and supports clients on geospatial initiatives. In parallel, he continues his academic research on AI-driven roadway feature extraction, intelligent mobility systems, and disaster debris assessment using remote sensing and geospatial analytics, with a particular focus on improving transportation resilience in rural communities. He can be contacted at rybantwi@gmail.com.