Improved Signalized Intersection Performance Using Computer Vision and Artificial Intelligence [Technical Summary]

Details

• Creators:
  Moomen, Milhan ; Codjoe, Julius
• Corporate Creators:
  Louisiana Transportation Research Center
• Subject/TRT Terms:
  • Intelligent transportation systems
  • Artificial intelligence
  • Deep learning
  • Tracking systems
  • Computer vision
  • Signalized intersections
  • Data analysis
  • Feasibility analysis
  • Performance measurement
• Publication/ Report Number:
  24-4SS
• Resource Type:
  Brief
• Geographical Coverage:
  United States
• Edition:
  Technical Summary
• Corporate Publisher:
  Louisiana Transportation Research Center
• Abstract:
  The objective of this project was to develop and evaluate artificial intelligence (AI)-based computer vision tools for intersection performance analysis. The specific objectives were to: • Develop a vehicle counting framework based on detection-tracking-counting pipeline for accurate turn-movement measurements; • Develop an integrated framework for vehicle queue length estimation, with associated traffic light state recognition; • Develop a vehicle-pedestrian interaction system to understand behaviors at intersections; and • Develop a user-friendly graphical interface (GUI) for practical applications by transportation engineers.
• Format:
  PDF
• Funding:
  DOTLT1000515
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/92001/dot_92001_DS1.pdf
• File Type:
  [PDF - 175.96 KB]
• Collection(s):
  Artificial Intelligence Deployment Center of Excellence (AI COE) ; US Transportation Collection
• Main Document Checksum:
  urn:sha-512:d1531d797441d9e1a4d0b9308206ffa2803d7a25c5d69acb5bda829727a0537450043fa38195ba1c597a152a3d7ab26547a83fe27034d229aace212a0aa96634