Artificial Intelligence–Aided Automated Detection of Railroad Trespassing

Large volumes of surveillance video data deployed in the railroad industry open many possibilities for detecting and preventing unsafe trespassing on railroad tracks. Monitoring these data, however, is highly time- and resource-consuming. In this article, authors describe an artificial intelligence (AI) algorithm that automatically detects trespassing events in real time. The system was tested on two different safety-critical scenarios: a grade crossing in Ashland, Virginia and two right-of-ways (ROWs) in Thomasville, North Carolina. The AI system was able to accurately detect trespasses in these locations. With this AI technology, it is possible to compile large data sets of trespassing events and provide useful insights into trespassing behavior to ultimately support risk mitigation decisions.

• Record URL:
  http://www.trb.org/Publications/Blurbs/179487.aspx
• Availability:
  • Find a library where document is available. Order URL: http://worldcat.org/issn/07386826
• Authors:
  • Zaman, Asim F
  • Ren, Baozhang
  • Liu, Xiang
• Publication Date: 2019-7

Language

• English

Media Info

• Media Type: Digital/other
• Features: Figures; Photos; References;
• Pagination: pp 30-35
• Serial:
  • TR News
  • Issue Number: 322
  • Publisher: Transportation Research Board
  • ISSN: 0738-6826

Subject/Index Terms

• TRT Terms: Algorithms; Artificial intelligence; Detection and identification systems; Fatalities; Railroad grade crossings; Railroad safety; Railroad tracks; Right of way (Land); Trespassers; Video
• Geographic Terms: Ashland (Virginia); Thomasville (North Carolina)
• Subject Areas: Pedestrians and Bicyclists; Railroads; Safety and Human Factors;

Filing Info

• Accession Number: 01713983
• Record Type: Publication
• Files: TRIS, TRB, ATRI
• Created Date: Aug 18 2019 9:55PM