Description

Addressing the issue of potholes is a primary concern for maintaining urban infrastructure. The research team has developed a prototype pothole management program. The program includes a mobile application and two machine learning models. The mobile app enables users to upload images of potholes, report relevant information, and provide driving directions to the pothole location. With the help of this application, the user can seamlessly capture images of the potholes, record pertinent information, and submit the data for necessary action. The mobile application is an essential tool in the Pothole Management Program (PMP), as it enhances the program's efficiency, effectiveness, and user experience. The program utilizes two machine learning models. The first model, Visual Geometry Group (VGG16), uses deep learning neural network technology to classify potholes with over 90% accuracy. The second machine learning model, You Only Look Once (YOLO), has been designed to detect and accurately mark potholes on submitted photos. Overall, this innovative pothole management program offers a comprehensive solution to help address the critical issue of potholes in urban areas.

Publication Date

7-2024

Publication Type

Report

Topic

Transportation Technology

Digital Object Identifier

10.31979/mti.2024.2306

MTI Project

2306

Keywords

Pavement condition, Pothole reporting, Tracking, Management

Disciplines

Computer Sciences | Data Science | Other Physical Sciences and Mathematics

Additional Files
2306-RB-Cheng-Pavement-Pothole-Reporting-Management (1).pdf (671 kB)
Research Brief
Download

Additional files available below

Share

COinS