CTR, in collaboration with the Texas Advanced Computing Center (TACC), is working on a project for the City of Austin that explores the use of data collected from traffic monitoring cameras to support traffic analysis and pedestrian behavior studies. This project leverages existing infrastructure for additional data collection, while minimizing the effort required to extract information from the collected data. The proposed methodology consists of an object recognition/tracking step, to identify relevant objects and store them in a structured database, and a data analysis step, which efficiently queries the dataset to extract the desired information.
While there are commercial products that use video analytics for various traffic management applications, they typically require the deployment of new hardware and are often designed to conduct a limited number of analyses to support real-time operations. Aside from the fact that it uses existing hardware, the approach under development by CTR is different from existing tools in that it efficiently stores the paths of recognized objects. The latter provides an opportunity to systematically save the information collected from lengthy video streams across multiple locations in Austin and query them over time as new questions rise and/or new analyses methodologies become available. Some of the applications explored so far include measuring traffic volumes and turning movement counts, and estimating changes in pedestrian patterns under different types of crosswalk treatment.