A hybrid system for an on-site automatic vehicle counting and classification

摘要

In the following paper authors present a novel solution to a task of an automatic vehicle counting and classification. The depicted, hybrid system is composed of a set of remote measurement units alongside a central computing node connected in a network. The developed software package deployed within this system is based on a digital video processing algorithms and contemporary neural network algorithms, for vehicle detection and detected object classification, respectively. Conducted measurements depict the performance of the designed system with respect to the agenda of the National Center for Research and Development RID project.