pacific nw transportation consortium region 10 Pactrans

Computer Vision for Pedestrian Traffic Monitoring

PI: Ahmed Abdel Rahim (UI), ahmed@uidaho.edu, ORCID: 0000-0001-9756-554X

Co PIs: Matthew Calhoun (UAA)

AMOUNT & MATCH: $50,000 federal from PacTrans; $50,000 federal Match

PERFORMANCE PERIOD: 8/16/2023 – 8/15/2025

STATUS: Active

CATEGORIES: Sensors, Mobility

UTC PROJECT DOCUMENTATION:

FINAL PROJECT REPORT: will be available once completed

PROJECT DATA: will be available once completed

DESCRIPTION:  Computer vision sensors provide several advantages over traditional traffic counting systems. They provide counts for more users: cars, trucks, bikes, and pedestrians, the speeds the vehicles are traveling, and the trajectory through space. The data is saved in a cloud-based system that is easy to query and download for further analysis. In addition, all of the information is recorded without keeping any personally, identifiable information to address the privacy concerns of jurisdictions and their citizens.

Six Numina Computer Vision sensors were purchased as part of a previous PacTrans project. Two sensors were installed in Boise, Idaho in November 2022. Two will be installed later this summer at different locations in Boise and two will be installed in Moscow, Idaho. The research team has previously investigated bicyclists riding under varying weather conditions. First, it sought to determine how the bike lane was used (or not used) during adverse weather. The hypothesis for this analysis was that bicyclists were riding in the street because the bike lane was left unplowed.

Data from these detection zones for the range between December 1st to February 6th was integrated with data for the same date range collected at the Boise Airport weather station. Counts were then retrieved for each of these days and compiled for the Bike Lane and the Street zone. The preliminary analysis confirmed our hypothesis that cyclists would ride in the street on days of adverse weather. Next, it used the same data to create prediction models for bicycle volumes.

The preliminary analysis was a success in terms of assessing the capabilities of the equipment. However, we only have three months of data from two locations. This project will provide the opportunity to collect a full year of data at the first two locations in Boise and provide the opportunity to explore other case study data for different situations in another community.