Evaluation of Near-Miss Crashes Using a Video-Based Tool

Vision-based trajectory data provide useful information on the trajectories of roadway users (vehicles, pedestrians, bicycles) that could be used for analyzing roadway safety and users’ interactions. Several companies offer vision-based software that promise to identify near-miss crashes between vehicles, pedestrians and vehicles, or bicycles and vehicles, by estimating surrogate measures for near-misses through the trajectory data. However, the accuracy of these methods in terms of predicting near-misses has not been evaluated. Therefore, the objective of this research is to install a commonly-used video-based software, and evaluate its accuracy of predicting near-misses at a busy signalized intersection in the Lawrence, KS area. The main objective of this research is to evaluate a video-based tool offered by “Brisk Synergies” in terms of its accuracy of predicting near-misses at signalized intersections. The findings of this study will help KDOT and local transportation partners to decide whether to invest or not in video-based tools at signalized intersections.


    • English


    • Status: Active
    • Funding: $71500
    • Contract Numbers:

      K-TRAN: KU-21-4



    • Sponsor Organizations:

      Kansas Department of Transportation

      Eisenhower State Office Building
      700 SW Harrison Street
      Topeka, KS  United States  66603-3754
    • Performing Organizations:

      University of Kansas Center for Research, Incorporated

      2291 Irving Hill Drive, Campus West
      Lawrence, KS  United States  66045
    • Principal Investigators:

      Schrock, Steven

      Kondyli, Alexandra

    • Start Date: 20200701
    • Expected Completion Date: 20221231
    • Actual Completion Date: 0

    Subject/Index Terms

    Filing Info

    • Accession Number: 01760107
    • Record Type: Research project
    • Source Agency: Kansas Department of Transportation
    • Contract Numbers: K-TRAN: KU-21-4, RE-0816-01, C2171
    • Files: RIP, STATEDOT
    • Created Date: Dec 15 2020 3:29PM