Exploring Cost-effective Computer Vision Solutions for Smart Transportation Systems

The key accomplishments of the project are summarized as follows. The team conducted a literature review on computer vision for smart cities with a focus on transportation and summarized a resource list that lists publicly available traffic camera systems in the U.S. The team also established an automatic pipeline for data acquisition and developed a web-based tool that integrated the computer vision algorithms for the two selected applications. The urban work zone application (WorkZoneX) leverages 900+ traffic cameras in New York City (NYC) and provides real-time urban work zone identification, active work zone with workers detection, work zone size estimation, and traffic condition around the work zones. WorkZoneX achieved an average mAP of 74.1% across all work zone classes, an accuracy of 98.4% for scene identification, and an accuracy of up to 89.52% for size estimation. The safety risk index map application (SAFExMAP) provides a risk indicator scoring system with a map interface that leverages near-miss data gathered from in-vehicle cameras via computer vision. A positive spatial correlation was found between near misses and crashes for the study area. Both applications were optimized for web-based access and prototyped for real-world deployment. A cost estimation of deploying the two applications was provided. This project stands to facilitate the adoption of computer vision in smart cities, potentially positively impacting transportation planning and operations by providing cost-effective solutions to the industry.

• Record URL:
  https://cip.tandoncsmart.com/uploads/1693418444053-Gao_ComputerVisionSolutions_2023.pdf
• Record URL:
  https://rosap.ntl.bts.gov/view/dot/72518
• Supplemental Notes:
  • This document was sponsored by the U.S. Department of Transportation, University Transportation Centers Program.
• Corporate Authors:

  Connected Cities for Smart Mobility towards Accessible and Resilient Transportation Center (C2SMART)

  New York University
  Tandon School of Engineering
  Brooklyn, NY  United States 

  Office of the Assistant Secretary for Research and Technology

  University Transportation Centers Program
  Department of Transportation
  Washington, DC  United States  20590

  Federal Highway Administration

  Office of Research, Development, and Technology
  McLean, VA  United States 
• Authors:
  • Gao, Jingqin
  • 0000-0002-1718-2432
  • Ozbay, Kaan
  • 0000-0001-7909-6532
  • Xu, Chuan
  • 0000-0001-5727-4561
  • Zhang, Daniel
  • 0009-0007-3372-4565
  • Zuo, Fan
  • 0000-0002-6761-2808
  • Yang, Liu
  • 0009-0009-7929-3582
  • Hammami, Omar
  • 0000-0002-4727-408X
• Publication Date: 2023-5

Language

• English

Media Info

• Media Type: Digital/other
• Edition: Final Report
• Features: Appendices; Figures; Maps; Photos; References; Tables;
• Pagination: 114p

Subject/Index Terms

• TRT Terms: Computer vision; Data collection; Detection and identification systems; Near crashes; Risk assessment; Web applications; Work zone safety
• Subject Areas: Data and Information Technology; Highways; Safety and Human Factors;

Filing Info

• Accession Number: 01898586
• Record Type: Publication
• Contract Numbers: 69A3551747119
• Files: UTC, NTL, TRIS, ATRI, USDOT
• Created Date: Nov 9 2023 5:10PM