Developing a Cost-Effective Bus-to-Pedestrian Near-Miss Detection Method using Onboard Video Camera Data
PI: Yinhai Wang (UW), yinhai@uw.edu
Co PIs: Weibin Zhang (UW), Ruimin Ke (UW)
AMOUNT & MATCH: $46,000 from PacTrans; $46,000 Match
PERFORMANCE PERIOD: 12/15/2016 – 1/31/2018
STATUS: Completed
CATEGORIES: ITS, Transit, Big Data
RESEARCH PROJECT HOT SHEET: 
UTC PROJECT DOCUMENTATION:
FINAL PROJECT REPORT:
DESCRIPTION: The primary goal of the project is to develop a method that can detect bus-to-pedestrian near-miss events in order to support bus-to-pedestrian collision avoidance applications and provide additional data source for safety statistical analysis. Specifically, we will define indicators for near-miss events in videos taken by onboard front facing cameras, then develop a video processing framework to automatically detect and record near-misses. Given the fact that the northwest region has a large number of transit agencies and relies heavily on transit buses in terms of commuting, such near-miss detection method will definitely obtain additional information to enhance further research on bus-to-pedestrian safety studies in the region. Potentially, the proposed method can save cost for related agencies and improve both pedestrian and bus safety.
| DELIVERABLE | DUE DATE | DATE RECEIVED |
| Research Project Progress Report #1 | April 10, 2017 | April 15, 2017 |
| Research Project Progress Report #2 | October 10, 2017 | October 9, 2017 |
| No Cost Extension Request | November 30, 2017 | never received |
| Draft Report | November 30,2017 | June 16, 2018 |
| Final Project Report | January 31, 2018 |
