An Automated System for Large-Scale Intersection Marking Data Collection and Condition Assessment

Intersection markings play a vital role in providing road users with guidance and information. These markings gradually degrade because of vehicular traffic, rain, and/or snowplowing. Degraded markings can confuse drivers, leading to increased risk of traffic crashes. Timely and high-quality information of intersection markings helps make informed decisions in safety management and maintenance prioritization. However, current labor-intensive and high-cost data collection practice makes it very challenging to gather intersection data on a large scale. This IDEA project developed an automated system to intelligently detect and characterize intersection markings and to assess their degradation conditions with existing roadway Geographic information systems (GIS) data and aerial images. The system focuses on two types of markings at intersections - lane-use arrows and crosswalks - and can be extended to also cover other road markings. The seamless integration of spatial analytics and advanced computer vision techniques makes the proposed system truly cost-effective, scalable, and computationally efficient. The system harnesses emerging artificial intelligence techniques such as transfer learning and multi-task deep learning to enhance its robustness, accuracy, and computational efficiency. A data acquisition module was developed to automatically retrieve intersection locations from roadway GIS data in Virginia and capture corresponding aerial images on a large scale. Over 3,000 intersection images were captured and manually annotated. Marking images were synthesized from different environment settings, and the synthesized data was used in a transfer learning process to pre-train computer vision models for marking detection and characterization. Computer vision modules were developed to detect and classify lane-use arrows (85% average precision) and crosswalks (89% average precision), to reliably measure the marking sizes by calibrating detection bounding boxes, and to assess the degradation conditions of markings (91% overall accuracy for lane-use arrows and 83% for crosswalks). Data acquisition and computer vision modules developed were integrated and a graphical user interface (GUI) was built for the system. The developed system has been deployed as a web-based application so that it does not require powerful client computers and any users with internet connections can easily access it. The proposed system can fully automate the processes of marking data collection and condition assessment on a large scale with almost zero cost and short processing time Current data collection practices require state agencies to invest millions of dollars in contracting very time-consuming data collection services each year. By economically providing large-scale intersection marking data, the system can help state agencies in their data-driven safety management and maintenance prioritization for ensuring intersection safety. The final report is available.

• Record URL:
  http://apps.trb.org/cmsfeed/TRBNetProjectDisplay.asp?ProjectID=5059

Language

• English

Project

• Status: Completed
• Funding: $135000
• Contract Numbers:

  20-30/IDEA 225

• Sponsor Organizations:

  National Cooperative Highway Research Program

  Transportation Research Board
  500 Fifth Street, NW
  Washington, DC  United States  20001

  Federal Highway Administration

  1200 New Jersey Avenue, SE
  Washington, DC  United States  20590

  American Association of State Highway and Transportation Officials (AASHTO)

  444 North Capitol Street, NW
  Washington, DC  United States  20001
• Project Managers:

  Jawed, Inam

• Performing Organizations:

  Old Dominion University

  ,    
• Principal Investigators:

  Xie, Kun

• Start Date: 20210101
• Expected Completion Date: 0
• Actual Completion Date: 0

Subject/Index Terms

• TRT Terms: Automation; Computer vision; Data collection; Detection and identification systems; Geographic information systems; Intersections; Machine learning; Neural networks; Prototypes; Road markings
• Subject Areas: Data and Information Technology; Highways; Operations and Traffic Management;

Filing Info

• Accession Number: 01757155
• Record Type: Research project
• Source Agency: Transportation Research Board
• Contract Numbers: 20-30/IDEA 225
• Files: TRB, RIP
• Created Date: Nov 4 2020 11:13AM