Computer Vision For Automated Traffic Safety Assessment : A Machine Learning Approach

Traffic safety is a complex and important research area with the potential to save many lives in the future. This article introduces a conceptual Traffic Surveillance Vision Pipeline (TSVP) which has the potential to solve two key problems in traffic safety, namely the gathering of reliable and detailed road user statistics and taking advantage of surveillance infrastructure to guide and assist vehicles in real time. Research is performed to experiment with and improve upon computer vision models and methods in the different parts of the TSVP. This includes a fast and efficient method for identifying empty and occupied parking spaces, an object detector design that takes advantage of the properties of surveillance videos, a software framework that implements most of the TSVP using established computer vision methods, a camera calibration method for a Trinocular Linear Camera Array (TLCA), and a tracking method in pixel coordinates.